METHOD, APPARATUS AND NON- TRANSIENT STORAGE MEANS

专利摘要:
METHOD, AND APPARATUS. The present patent application primarily relates to specific methods for inferring the activity of one or more cell signaling pathway(s) in the tissue of an individual under medical evaluation, at least based on the expression level(s) of one or more target gene(s) of the cell signaling pathway(s) measured in an extracted tissue sample of medical material, an apparatus comprising a digital compressor configured to perform such methods and a means of non-transient storage storing instructions that are executable by a digital processing device to perform such methods.
公开号:BR112014000965B1
申请号:R112014000965-1
申请日:2012-07-19
公开日:2021-08-24
发明作者:Wilhelmus Franciscus Johannes Verhaegh；Anja Van De Stolpe；Hendrik Jan Van Ooijen；Kalyana Chakravarthi Dulla；Marcia Alves De India；Ralf Hoffmann
申请人:Koninklijke Philips N.V.；
IPC主号:

专利说明:

[0001] The subject matter described in this document refers to bioinformatics, genomic processing techniques, proteomic processing techniques, and related techniques.
[0002] Genomic and proteomic analyzes have substantial applications realized and potential promise for clinical application in medical fields such as oncology, where several cancers are known to be associated with specific combinations of mutations/genomic variations and/or high or low expression levels for specific genes, which play a role in cancer growth and evolution, eg cell proliferation and metastasis. For example, the Wnt signaling pathway affects the regulation of cell proliferation, and is highly regulated. The high activity of the Wnt pathway due to loss of regulation has been correlated with cancer, including malignant colon tumors. Although not limited to any specific theory of operation, dysregulation of the Wnt pathway in malignant colon cells is believed to lead to high activity of the Wnt pathway which, in turn, causes cell proliferation of the malignant colon cells, i.e. , spread of colon cancer. On the other hand, abnormally low pathway activity may also be of interest, for example, in the case of osteoporosis.
[0003] Technologies to acquire genomic and proteomic data have become readily available in clinical practice. For example, microarray measurements are routinely employed to assess gene expression levels, protein levels, methylation, and so on. Automated gene sequencing allows cost-effective identification of genetic variations in DNA and mRNA. Quantitative assessment of mRNA levels during sequencing holds promise for yet another clinical tool for assessing gene expression levels.
[0004] Despite these (or perhaps because of these) advances, the clinical application of genomic and proteomic analyzes faces a major obstacle - data overload. For example, the number of identifiable mutations in a single clinical sample can be in the hundreds or thousands or more. Most of these mutations are called bystander mutations with no specific contribution to cancer growth, and only a few contribute to cancer growth and functional evolution, and these present the targets for effective treatment. A single microarray can generate gene expression levels for tens of thousands of genes. Processing these large amounts of data to identify useful information, such as in applying the right therapy choice, is difficult.
[0005] One approach is to limit the analysis to a few canonical or standardized tests, such as tests approved by the U.S. Food and Drug Administration (FDA). In such an approach, a specific indicator or combination of indicators (eg mutations, and/or high or low gene expression levels is detected to test “positive for the indicated disease condition (eg a specific type of cancer). Canonical testing is confirmed by clinical studies that have shown strong correlation with disease status or treatment efficacy. This approach is only useful for clinical conditions for which a canonical test has been developed, eg, specific diagnosis of a disease , or predicting a response to a drug in a specific type of cancer at a specific stage, and is also rigid as it is only applicable for canonical conditions.
[0006] Another approach is based on the identification of functionally related groups of genomic or proteomic indicators. For example, the Wnt pathway comprises a cascade of proteomic reactions. Major components of this chain include (but are not limited to) the binding of the Wnt signaling protein to a crimped cell surface receptor, which causes activation of proteins from the disheveled family of proteins which, in turn, impacts the level of transcription agents such as β-catenin/TCF4-based protein complexes in the cell nucleus. These transcription agents, in turn, control the transcription of target mRNA molecules which, in turn, are translated into target proteins of the Wnt pathway. Clinical studies have shown some correlations between Wnt pathway regulatory proteins and Wnt pathway activity.
[0007] However, the application of such clinical study results to the diagnosis and clinical assessment of a specific patient is difficult due to the complexity of signaling pathways, for example, the Wnt pathway. As a simple example, measuring the expression level of a protein that is “upstream” in the Wnt pathway may not detect the abnormal behavior of a protein that is “downstream” in the Wnt pathway. The Wnt pathway is believed to include several feedback mechanisms, and the simplified concept of “upstream” and “downstream” may not be applicable for a substantial portion of the Wnt pathway; more generally, abnormal behavior in one portion of the protein cascade comprising the Wnt pathway may have more or less effect on other portions of the protein cascade, and on the activity of the Wnt pathway as a whole. Furthermore, in some clinical studies, protein expression levels for regulatory proteins of the signaling cascade are assessed by measuring the mRNA expression levels of the genes that code for the regulatory proteins. This is an indirect measurement that can inaccurately assess the level of regulatory protein expression, and hardly reflects the amount of active proteins (after a specific post-translational modification such as phosphorylation).
[0008] The main problem underlying the present invention was, thus, to provide adequate methods and means to perform genomic and, respectively, proteomic analyses. Specific aspects of the underlying problem, as well as additional objections in connection with the present invention, become apparent upon studying the description, the examples provided in this document, and, in particular, upon studying the appended claims.
[0009] The present invention provides new and improved methods and apparatus as disclosed in this document.
[0010] According to a main aspect of the present invention, the above problem is solved by a specific method to assess cell signaling pathway activity using probabilistic target gene expression modeling, namely, a method comprising:
[0011] the inference of the activity of one or more cell signaling pathway(s) in the tissue of a medical material, at least based on the level(s) of expression (in particular, on the mRNA and/or level of protein) of one or more target gene(s) of the cell signaling pathway(s) measured in a sample extracted from the tissue of medical material, where the inference comprises:
[0012] the inference of the activity of the cell signaling pathway(s) in the tissue of medical material by evaluating at least a portion of a probabilistic model, preferably a Bayesian network, representing the pathway(s) signaling pathway(s) for a set of inputs including at least the expression level(s) of one or more target genes of the cell signaling pathway(s) measured in the extracted tissue sample of the material doctor;
[0013] the estimation of a tissue level in medical material of at least one transcription factor (TF) element, the at least one TF element controlling the transcription of one or more target gene(s) cell signaling pathway(s), the estimate being based, at least in part, on conditional probabilities relating to at least one TF element and the expression level(s) of the one or more gene(s) ) target of the cell signaling pathway(s) measured in the sample extracted from the tissue of the medical material; and
[0014] the inference of the activity of the cell signaling pathway(s) based on the estimated level in the tissue sample of the transcription factor; and
[0015] the determination of the cell signaling pathway(s) to be/are or not operating abnormally in the tissue of the medical material based on the inferred activity of the cellular signaling pathway(s) in the tissue of the medical material ;
[0016] in which the inference is performed by a digital processing device using the probabilistic model of cell signaling pathway(s).
The "target gene(s)" can be "direct target genes" and/or "indirect target genes" (as described in this document).
[0018] Preferably, the inference comprises the estimation of a tissue level of medical material of at least one transcription factor (TF) element represented by a TF node of the probabilistic model, the TF element controlling the transcription of one or more gene( s) target of the cell signaling pathway(s), the estimate being based, at least in part, on conditional probabilities of the probabilistic model related to the TF node and nodes in the probabilistic model representing the one or more target gene(s) of the cell signaling pathway(s) measured in the tissue sample taken from the medical material.
[0019] The probabilistic model can be a Bayesian network model. Thus, according to a preferred embodiment, the inference is performed using a Bayesian network comprising nodes representing information about the signaling pathway(s) and conditional probability relations between connected nodes of the Bayesian network.
The cell signaling pathway(s) may be a Wnt pathway, an ER (Estrogen Receptor) pathway, an AR (Androgen Receptor) pathway and/or a Hedgehog pathway. Thus, according to a preferred embodiment, the cell signaling pathway(s) comprises a Wnt pathway, an ER pathway, an AR pathway and/or a Hedgehog pathway.
[0021] Particularly suitable target genes are described in the following text passages, as well as in the examples below (see, for example, Tables 1 to 9).
[0022] Thus, according to a preferred embodiment, the target gene(s) is/are selected from the group comprising or consisting of target genes listed in Table 1 or Table 6 (for the via the Wnt pathway), target genes listed in Table 2, Table 5 or Table 7 (for ER pathway), target genes listed in Table 3 or Table 8 (for Hedgehog pathway) and target genes listed in Table 4 or Table 9 (for via the AR).
[0023] Particularly preferred is a method where the inference comprises:
[0024] the inference of the activity of a Wnt pathway in the tissue of medical material based on at least expression levels of one or more, preferably at least three, target gene(s) of the Wnt pathway measured in the sample extracted from the tissue of the medical material selected from the group comprising or consisting of: KIAA1199, AXIN2, RNF43, TBX3, TDGF1, SOX9, ASCL2, IL8, SP5, ZNRF3, KLF6, CCND1, DEFA6 and FZD7.
[0025] Also preferred is a method wherein the inference is further based on expression levels of at least one target gene of the Wnt pathway measured in the extracted tissue sample of medical material selected from the group comprising or consisting of: NKD1 , OAT, FAT1, LEF1, GLUL, REG1B, TCF7L2, COL18A1, BMP7, SLC1A2, ADRA2C, PPARG, DKK1, HNF1A and LECT2.
[0026] Particularly preferred is a method where the inference (also) comprises:
[0027] the inference of the activity of an ER pathway in tissue from medical material based on at least expression levels of one or more, preferably at least three, target gene(s) of the ER pathway measured in the extracted tissue sample of medical material selected from the group comprising or consisting of: CDH26, SGK3, PGR, GREB1, CA12, XBP1, CELSR2, WISP2, DSCAM, ERBB2, CTSD, TFF1 and NRIP1.
[0028] Also preferred is a method in which the inference is additionally based on expression levels of at least one target gene of the ER pathway measured in the extracted tissue sample of medical material selected from the group comprising or consisting of: AP1B1 , ATP5J, COL18A1, COX7A2L, EBAG9, ESR1, HSPB1, IGFBP4, KRT19, MYC, NDUFV3, PISD, PREDM15, PTMA, RARA, SOD1 and TRIM25.
[0029] A method in which the inference (also)comprises
[0030] the inference of the activity of a Hedgehog pathway in the tissue of medical material based on at least expression levels of one or more, preferably at least three, target gene(s) of the Hedgehog pathway measured in the sample extracted from the tissue of the material physician selected from the group that comprises or consists of: GLI1, PTCH1, PTCH2, IGFBP6, SPP1, CCND2, FST, FOXL1, CFLAR, TSC22D1, RAB34, S100A9, S100A7, MYCN, FOXM1, GLI3, TCEA2, FYN and CTSL1,
[0031] is also preferred.
[0032] Also preferred is a method in which the inference is additionally based on expression levels of at least one target gene of the Hedgehog pathway measured in the extracted tissue sample of medical material selected from the group comprising or consisting of: BCL2, FOXA2, FOXF1, H19, HHIP, IL1R2, JAG2, JUP, MIF, MYLK, NKX2.2, NKX2.8, PITRM1 and TOM1.
[0033] A method in which the inference (also)comprises
[0034] the inference of the activity of an AR pathway in tissue from medical material based on at least expression levels of one or more, preferably at least three, AR pathway target gene(s) measured in the extracted tissue sample of the medical material selected from the group that comprises or consists of: KLK2, PMEPA1, TMPRSS2, NKX3_1,ABCC4, KLK3, FKBP5, ELL2, UGT2B15, DHCR24, PPAP2A, NDRG1, LRIG1, CREB3L4, LCP1, GUCY1A3, AR and EAF2,
[0035] is also preferred.
[0036] Also preferred is a method in which the inference is additionally based on expression levels of at least one target gene of the AR pathway measured in the extracted tissue sample of medical material selected from the group comprising or consisting of: APP , NTS, PLAU, CDKN1A, DRG1, FGF8, IGF1, PRKACB, PTPN1, SGK1 and TACC2.
[0037] Another aspect of the present invention relates to a method (as described herein), further comprising:
[0038] the prescription of a drug for medical material that corrects the abnormal operation of the cell signaling pathway(s);
[0039] in which the recommendation is made only if it is determined that the cell signaling pathway(s) is/are operating abnormally in the tissue of the medical material based on the inferred activity of the pathway(s) of cell signaling.
[0040] The present invention also relates to a method (as described in this document) comprising:
[0041] the inference of the activity of a Wnt pathway in tissue from a medical material at least based on expression levels of two, three or more target genes from a set of target genes of the Wnt pathway measured in a sample extracted from the medical material tissue
[0042] and/or
[0043] the inference of the activity of an ER pathway in tissue from a medical material at least based on expression levels of two, three or more target genes from a set of ER pathway target genes measured in a sample extracted from the tissue of medical supplies
[0044] and/or
[0045] the inference of the activity of a Hedgehog pathway in tissue from a medical material at least based on expression levels of two, three or more target genes from a set of target genes of the Hedgehog pathway measured in a sample extracted from the medical material tissue
[0046] and/or
[0047] the inference of the activity of an AR pathway in tissue from a medical material at least based on expression levels of two, three or more target genes from a set of AR pathway target genes measured in a sample extracted from tissue of medical supplies.
[0048] Preferably,
[0049] the set of target genes of the Wnt pathway includes at least nine, preferably all target genes selected from the group comprising or consisting of: KIAA1199, AXIN2, RNF43, TBX3, TDGF1, SOX9, ASCL2, IL8, SP5 , ZNRF3, KLF6, CCND1, DEFA6 and FZD7,
[0050] and/or
[0051] the set of ER pathway target genes includes at least nine, preferably all target genes selected from the group comprising or consisting of: CDH26, SGK3, PGR, GREB1, CA12, XBP1, CELSR2, WISP2, DSCAM, ERBB2, CTSD, TFF1 and NRIP1,
[0052] and/or
[0053] the set of target genes of the Hedgehog pathway includes at least nine, preferably all target genes selected from the group comprising or consisting of: GLI1, PTCH1, PTCH2, IGFBP6, SPP1, CCND2, FST, FOXL1, CFLAR , TSC22D1, RAB34, S100A9, S100A7, MYCN, FOXM1, GLI3, TCEA2, FYNe CTSL1,
[0054] and/or
[0055] the set of target genes of the AR pathway includes at least nine, preferably all target genes selected from the group comprising or consisting of: KLK2, PMEPA1, TMPRSS2, NKX3_1, ABCC4, KLK3, FKBP5, ELL2, UGT2B15, DHCR24, PPAP2A, NDRG1, LRIG1, CREB3L4, LCP1, GUCY1A3, AR and EAF2.
[0056] A method in which
[0057] the set of target genes of the W pathway further includes at least one target gene selected from the group that comprises or consists of: NKD1, OAT, FAT1, LEF1, GLUL, REG1B, TCF7L2, COL18A1, BMP7, SLC1A2, ADRA2C, PPARG, DKK1, HNF1A and LECT2,
[0058] and/or
[0059] the set of ER pathway target genes additionally includes at least one target gene selected from the group comprising or consisting of: AP1B1, ATP5J, COL18A1, COX7A2L, EBAG9, ESR1, HSPB1, IGFBP4, KRT19, MYC , NDUFV3, PISD, PREDM15, PTMA, RARA, SOD1 and TRIM25,
[0060] and/or
[0061] the set of target genes of the Hedgehog pathway additionally includes at least one target gene selected from the group that comprises or consists of: BCL2, FOXA2, FOXF1, H19, HHIP, IL1R2, JAG2, JUP, MIF, MYLK, NKX2.2, NKX2.8, PITRM1 and TOM1,
[0062] and/or
[0063] the AR pathway target gene set additionally includes at least one target gene selected from the group comprising or consisting of: APP, NTS, PLAU, CDKN1A, DRG1, FGF8, IGF1, PRKACB, PTPN1, SGK1 and TACC2,
[0064] is particularly preferred.
[0065] The sample(s) to be used, according to the present invention may be, for example, a sample obtained from a breast lesion, or from a colon of a medical material known to, or suspected of having colon cancer, or a liver from medical material known to, or suspected of, having liver cancer, or so forth, preferably through a biopsy or other procedure sample extraction procedure. The tissue from which a sample is taken may also be metastatic tissue, for example, tissue (suspected to be) malignant originating from the colon, breast, liver, or other organ that has spread outside the colon, breast, liver, or other organ. In some cases, the tissue sample can be circulating tumor cells, that is, tumor cells that have entered the bloodstream and can be extracted like the extracted tissue sample using appropriate isolation techniques.
[0066] Another disclosed aspect of the present invention relates to the use of a non-transient storage medium, as described herein, or a computer program, as described in this document, for specific diagnosis of a disease or prediction of the response to a drug in a specific type of cancer at a specific stage.
[0067] According to another disclosed aspect, an apparatus comprising a digital processor configured to carry out a method according to the invention as described in this document.
[0068] According to another disclosed aspect, a non-transient storage medium stores instructions that are executable by a digital processing device to perform a method according to the invention as described in this document. The non-transient storage medium may be a computer-readable storage medium, such as a hard disk or other magnetic storage medium, an optical disk or other optical storage medium, a random access memory (RAM), memory only. read (ROM), flash memory, or other electronic storage medium, a network server, and so on. The digital processing device can be a portable device (for example, a personal data assistant or smartphone), a notebook computer, a desktop computer, a computer or tablet device, a remote network server, and so on.
[0069] According to another disclosed aspect, a computer program comprises program code means for causing a digital processing device to perform a method according to the invention as described in this document. The digital processing device can be a portable device (for example, a personal data assistant or smartphone), a notebook computer, a desktop computer, a computer or tablet device, a remote network server, and so on.
[0070] One advantage lies in a clinical decision support system (CDS) providing clinical recommendations based on probabilistic analyzes of one or more cell signaling pathway(s), for example, using a Wnt pathway Bayesian network model, an ER pathway, an AR pathway, and/or a Hedgehog pathway.
[0071] Another advantage lies in the better assessment of cell signaling pathway activity that is less susceptible to errors.
[0072] Another advantage lies in the provision of a CDS system recommending targeted treatment for loss of regulation of a cell signaling pathway.
[0073] Another advantage lies in the provision of a CDS system that is designed to detect loss of regulation for a specific cell signaling pathway, such as an Ent pathway, an ER pathway, an AR pathway or a Hedgehog pathway, and it is readily adapted to provide recommendations for different types of cancer originated by this specific cell designalization pathway.
[0074] The present invention as described in this document can, for example, be advantageously used in connection with
[0075] diagnosis based on predicted (inferred) activity;
[0076] prediction based on predicted (inferred) activity;
[0077] drug prescription based on predicted (inferred) activity;
[0078] prediction of drug efficacy based on predicted (inferred) activity;
[0079] prediction of adverse effects based on predicted (inferred) activity;
[0080] drug efficacy monitoring;
[0081] drug development;
[0082] test development;
[0083] route search;
[0084] cancer staging;
[0085] enrolling the subject in a clinical trial based on predicted (inferred) activity;
[0086] selection of the subsequent test to be performed, and/or;
[0087] selection of associated diagnostic tests.
[0088] Additional advantages will become apparent to those skilled in the art after reading and understanding the appended figures, the following description and, in particular, after reading the detailed examples provided in this document below.
[0089] Figure 1 shows a simple Bayesian network representing part of a cell signaling pathway. The cell signaling pathway is symbolized by a transcription factor (TF) complex and target genes produced as a result of the transcription factor complex being present. The probabilistic relationship between the TF element and a target gene in the case of binary discretization can be represented by a conditional probability table as illustrated in the diagram.
[0090] Figure 2 shows an illustrative Bayesian network describing a hypothetical cell signaling pathway. Both upstream proteins and downstream target mRNA nodes are illustrated in the diagram. Upstream proteins serve as input to the transcription factor complex, while target mRNAs are the output nodes of the transcription factor complex.
[0091] Figure 3 shows an illustrative example of a Bayesian network representation of a single cell signaling pathway with multiple transcription factor complexes or multiple cell signaling pathways with its own transcription factor complex combined in a Bayesian network or a combination of them.
[0092] Figure 4 shows an example of a Bayesian network illustrating a simple representation of a cell signaling pathway similar to Figure 1. Now, additional nodes have been attached to represent the translation of target mRNA into target proteins.
[0093] Figure 5 shows an illustration of a Bayesian network illustrating another simple representation of a cell signaling pathway. The pathway is represented using the transcription factor complex and its target protein levels.
[0094] Figure 6 shows the illustrative Bayesian network of Figure 1 with an additional layer of nodes representing the sets of probes on a microarray chip connecting the probe intensities to the corresponding target mRNA levels.
[0095] Figure 7 shows an illustrative example of a variant embodiment of the Bayesian network of Figure 1 that includes nodes representing methylation and copy number variations as examples for additional information nodes for, in this specific example, any of the mRNA levels target included.
[0096] Figure 8 shows a Wnt pathway activity of the Bayesian network and closest centroid method as described in this document on a colon sample dataset (GSE20916).
[0097] Figure 9 shows a Bayesian network Wnt pathway activity and closest centroid method as described in this document on a colon sample dataset (GSE4183).
[0098] Figure 10 shows a Wnt pathway activity of the Bayesian network and closest centroid method as described in this document on a colon sample dataset (GSE15960).
[0099] Figure 11 shows a Wnt pathway activity of the Bayesian network and closest centroid method as described in this document on a breast cancer sample dataset (GSE12777).
[0100] Figure 12 shows a Wnt pathway activity of the Bayesian network and closest centroid method as described in this document in a breast cancer sample dataset (GSE21653).
[0101] Figure 13 shows a Wnt pathway activity of the Bayesian network and closest centroid method as described in this document in a liver cancer sample dataset (GSE9843).
[0102] Figure 14 shows a Wnt pathway activity of the Bayesian network and using target genes from the curated list of evidence compared to target genes from the broad list of literature as described in this document in a sample dataset of colon (GSE20916).
[0103] Figure 15 shows a Wnt pathway activity of the Bayesian network and using target genes from the curated list of evidence compared to target genes from the broad list of literature as described in this document in a sample dataset of colon (GSE4183).
[0104] Figure 16 shows a Wnt pathway activity of the Bayesian network and using target genes from the curated list of evidence compared to target genes from the broad list of literature as described in this document in a sample dataset of colon (GSE15960).
[0105] Figure 17 shows a Wnt pathway activity of the Bayesian network and using target genes from the curated list of evidence compared to target genes from the broad list of literature as described in this document in a sample dataset of breast cancer (GSE12777).
[0106] Figure 18 shows a Wnt pathway activity of the Bayesian network and using target genes from the curated list of evidence compared to target genes from the broad list of literature as described in this document in a sample dataset of liver cancer (GSE9843).
[0107] Figure 19 shows a Wnt pathway activity of the Bayesian network and using target genes from the curated list of evidence compared to target genes from the broad list of literature as described in this document in a sample dataset of medulloblastoma (GSE10327).
[0108] Figure 20 schematically shows a clinical decision support system (CDS) configured to assess one or more cellular signaling pathway(s) as disclosed in this document (example shown for the Wnt pathway).
[0109] Figure 21 shows a predicted Wnt pathway activity in GSE4183 colon samples.
[0110] Figure 22 shows a predicted Wnt pathway activity in GSE10327 medulloblastoma samples.
[0111] Figure 23 shows a predicted Wnt pathway activity in liver cancer samples from GSE9843.
[0112] Figure 24 shows a predicted Wnt pathway activity in GSE12777 breast cancer cell lines.
[0113] Figure 25 shows a predicted ER pathway activity in GSE12777 breast cancer cell lines.
[0114] Figure 26 shows a predicted ER pathway activity in breast cancer samples of GSE12276.
[0115] Figure 27 shows a predicted ER pathway activity in GSE36133 cancer cell lines.
[0116] Figure 28 shows a predicted viaHedgehog activity in GSE34211 cancer cell lines.
[0117] Figure 29 shows a predicted viaHedgehog activity in GSE10327 medulloblastoma samples.
[0118] Figure 30 shows a predicted viaHedgehog activity in breast cancer samples of GSE12276.
[0119] Figure 31 shows a predicted ER pathway activity in MCF7 and Tamoxifen resistant cell lines of GSE21618.
[0120] Figure 32 shows a predicted ER pathway activity in a time series of MCF7 cell line samples stimulated with estrogen from GSE11324.
[0121] Figure 33 shows the activity of viaWnt, ER and Hedgehog in luminal A samples of GSE12276.
[0122] Figure 34 shows the activity of viaWnt, ER and Hedgehog in basal samples of GSE12276.
[0123] Figure 35 shows predicted viaWnt activity in colon samples of GSE20916.
[0124] Figure 36 shows a predicted ER pathway activity in MCF7 cell lines stimulated with estrogen (E2) or a negative control (EtOH) (GSE9253).
[0125] Figure 37 shows Kaplan-Meier survival curves for patients from the GSE12276 dataset grouped according to pathway activity.
[0126] Figure 38 shows predicted AR pathway activity in LNCaP cell lines treated with different GSE7708 treatment regimens.
[0127] Figure 39 shows predicted AR pathway activity in prostate cancer samples from GSE17951.
[0128] Figure 40 shows a predicted AR pathway activity in breast cancer samples of GSE12276.
[0129] Figure 41 shows an AR pathway activity in the GSE36133 dataset containing cell line samples representing various types of cancer.
[0130] Figure 42 shows an AR pathway activity in the GSE34211 dataset containing cell line samples representing various types of cancer.
[0131] The following examples merely illustrate particularly preferred methods and selected aspects in connection therewith. The teachings provided in this document can be used to construct various tests and/or kits, for example, to detect, predict and/or diagnose the abnormal activity of one or more cell signaling pathways. In addition, after using the methods as described in this document, drug prescription can be advantageously guided, drug prediction and monitoring of drug efficacy (and/or adverse effects) can be done, drug resistance can be predicted and monitored, for example, to select subsequent test(s) to be performed (such as an associated diagnostic test). The following examples are not to be interpreted as limiting the scope of the present invention.EXAMPLE 1: BAYESIAN NETWORK CONSTRUCTION
[0132] As revealed in this document, by building a probabilistic model (eg, the illustrative Bayesian model shown in Figure 6) and incorporating probabilistic relationships between expression levels of a number of different target genes and cell signaling pathway activity, such a model can be used to determine cell signaling pathway activity with a high degree of accuracy. In addition, the probabilistic model can be readily updated to incorporate additional knowledge gained from further clinical studies by adjusting conditional probabilities and/or adding new nodes to the model to represent additional sources of information. In this way, the probabilistic model can be updated as appropriate to incorporate the latest medical knowledge.
[0133] One of the simplest Bayesian network models to represent a cell signaling pathway would be a two-level model including the transcription factor element and associated target genes (see Figure 1). The transcription factor complex element is a representation of the transcription factor complex level. The protein level of the transcription factor element is connected to a series of mRNA levels of the transcription factor target genes (in this exemplary Bayesian network, only three target genes are illustrated, which are known to be expressed in tissue in case the transcription factor is available). It should be understood that many, most, or all target genes of the pathway (in the case of the Wnt, ER, Hedgehog and Ar pathways, particularly the genes mentioned in Table 1, Table 2, Table 3 and Table 4, respectively) are similarly regulated by the TF element. The relationships between the TF element level and the mRNA levels of the target genes are modeled in the Bayesian network by the borders. For each of the target genes, a conditional probability distribution specifies how the mRNA level of the gene depends on the level of the TF element.
[0134] TF element levels and target genes can be represented in several ways. One option is to use a binary discretization, in “absent” and “present” states for the TF element, and “low” and “high” for the mRNA level of the target gene (see Figure 1). The probabilistic relationship between the TF element and a target gene can then be represented by a conditional probability table (as indicated in the same figure). Instead of a binary discretization, levels can also be represented as continuous level values, or as quantized values having three or more quantization levels (eg, "low", "normal", and "high" for target genes ).
[0135] The previous illustration of a simple Bayesian network is just an illustrative realization of the Bayesian network model (Figure 1). In general, a Bayesian network model comprises an acyclic directed graph comprising nodes connected by edges. Each node represents an item of information pertaining to the pathway in question (or, more generally, to the cell signaling pathway). The pathway element nodes each represent a genomic or proteomic element of the cell signaling pathway. As an illustrative example, a pathway element node may represent one of, but not limited to: a protein, a protein complex, an mRNA molecule transcribed from a target cell signaling pathway gene, a methylated gene, a phosphorylated protein, a complex of phosphorylated protein, and so on. As discussed later in this document, several other types of nodes, but not limited to the examples given, can be included in the Bayesian network to represent other types of information, such as specific measurement data elements, occurrences of variation, and so on.
[0136] Additional “upstream” levels representing pathway regulatory proteins (in active or inactive state) are typically added if knowledge of the level of such a protein can be evidenced to determine the clinical decision support recommendation. For example, the inclusion of elementary proteins to the transcription factor or essential proteins upstream of the transcription factor in the Bayesian network (see Figure 2) could be useful if a drug is available that specifically targets these proteins, rather than the pathway as a whole. . The transcription factor (TF) element is believed to be a protein complex (that is, a combination of proteins linked together in a specific structure that performs the function of regulating transcription of target genes) in most pathways. signage. For other pathways, the TF element can be a single protein. Furthermore, signaling pathways can exert their activities through more than one transcription factor, resulting in a more complex Bayesian network with several transcription factors feeding the target gene(s) (see Figure 3 for an illustration hypothesis of several transcription factor elements influencing target gene transcription). Such a multiple transcription factor Bayesian network may also be the result of a combination of pathways combined in a Bayesian network.
[0137] Additional information nodes further downstream from the target genes can be included in the Bayesian network as well. An illustrative example of this is the translation of target gene mRNA into proteins (Figure 4) or target gene protein level nodes as surrogate node for target gene mRNA level (Figure 5). Target gene mRNA molecules are translated by interacting with ribosome molecules to form proteins corresponding to mRNA molecules and corresponding to target genes. This is the expression of target genes at the protein level. Protein level measurement through, among others, eg mass spectrometry, immunohistochemistry, gel electrophoresis techniques, can act as evidence for these target protein levels.
[0138] The expression level of a target gene can be computed based on the measured intensity of corresponding probe sets of a microarray, for example, by averaging or by other means of other techniques (eg, RNA sequencing) . In some embodiments, this computation is integrated into the Bayesian network, by extending the Bayesian network with one node for each set of probes that is used and including an edge that goes to each of these “measurement” nodes from the corresponding target gene node , as described in this document with reference to Figure 6.
[0139] The probabilistic model may optionally also incorporate additional genomic information, such as information about mutations, copy number variations, gene expression, methylation, translocation information, and so on, which alter the genomic sequences that are related to the cascade of signaling pathway to infer pathway activity and locate the defect in the Wnt pathway causing aberrant functioning (activation or inactivity), as described in illustrative reference to Figure 7 for the illustrative case of methylation and copy number data. However, it should be understood that other types of information concerning the target gene are similarly translated into information nodes. Such genomic information may be available through, among others, RNA sequencing and SNP analysis.
[0140] Furthermore, it should be understood that, although the examples described later in this document referring to the Wnt, ER, AR and Hedgehog pathways are provided as illustrative examples, the approaches to cell signaling pathway analysis disclosed in this document are readily applied to other cell signaling pathways in addition to these pathways, such as intracellular signaling pathways with cell membrane receptors (eg, the Slot, HER2/PI3K, TGFbeta, EGF, VEGF, and TNF-NFkappaB signaling pathways) and pathways of intracellular signaling with receptors within the cell (eg, cell signaling pathways for progesterone, retinoic acid, and vitamin D).EXAMPLE 2: COMPARISON OF MACHINE LEARNING METHODS
[0141] Here, the performance of two machine learning techniques is compared to each other with the Wnt pathway taken as an example case: the prediction of Wnt activity by means of a closer centroid method is compared with the choice method according to the present invention, which, for example, uses a Bayesian network.
[0142] As discussed above, the Bayesian network approach was selected based on its advantages that lie in the probabilistic approach being able to incorporate the available information in "soft" form, for example, percentages of study subjects exhibiting evidentiary characteristics, or “hard”, using conditional probabilistic relationships. In addition, the probabilistic model also allows information to be incorporated based on partial (rather than complete) knowledge of the underlying cell signaling pathway, again through the use of conditional probability tables.
[0143] Here, it is shown that the inventors added value in the way they included known biological properties and the availability of soft evidence using a Bayesian network compared to other machine learning methods, eg closer centroid classification, a well known method. The closest centroid classification is a machine learning method where, for each class of training samples, a mean profile (= centroid) is computed, and then, for a sample to be classified, the label is predicted based on on the nearest centroid (the label of the nearest centroid is then the forecast result). The two centroids are calculated from the same probeset list used in the Bayesian network, and for the 'Wnt on' and 'Wnt off' centroid they are based on adenoma samples and normal colon samples, respectively, from the same data processed from fRMA of GSE8671. The log2 ratio of the two Euclidean distances between a sample and the two centroids was later used to classify samples from different datasets to infer sample classification. This means that a log2 ratio of 0 corresponds to an equal distance from the sample to the two centroids, a value > 0 corresponds to a sample classified as active Wnt signaling, while a value < 0 corresponds to an identified sample having a pathway Wnt flag inactive.
[0144] The Bayesian network was built similar to Figure 6 and the procedure described in this document. Analogous to this description of the Wnt Bayesian network, the conditional probability tables of the edges between probesets and their respective genes were trained using fRMA processed data from 32 normal colon samples and 32 adenoma samples from Gene dataset GSE8671 Expression Omnibus (accessible at http://www.ncbi.nlm.nih.gov/geo/, last accessed July 13, 2011). The trained Bayesian network was then tested on several data sets to infer the probability P(Wnt On) that the Wnt pathway is “on”, ie active, which is taken equal to the inferred probability that the transcription complex of the Wnt pathway is “present”.
[0145] The trained Bayesian network and the closest centroid model were then tested on several microarray datasets processed by fRMA to infer the probability that the Wnt pathway is “on”, measured by P(Wnt On) and log2 relation of distances. Summaries of the results of the Bayesian network and the closest centroid model are shown in Figures 8 through 13. The reader should note that the output metrics of the two methods are not a one-to-one relationship, however, the sign and magnitude The relative output metrics within a method are comparable.
[0146] The vast majority of colon (cancer) samples (GSE20916, GSE4183) are classified equally between the active and inactive Wnt pathway, except for GSE15960 which had a high fraction of samples mislabeled negative in the nearest centroid method (false negative). This perception of a higher fraction of false negatives is maintained in other cancers as well. This is especially true for breast cancer (GSE12777, GSE21653) and liver cancer (GSE9843) samples; with few exceptions, all specimens are predicted to have an inactive Wnt pathway, which is known to be incorrect in the case of basal type breast cancer and CTNNB1 liver cancer specimens. In some cases, evident, for example, in GSE15960, the classification could be corrected by decreasing and increasing the nearest centroid classification threshold. The idea behind this would be that the threshold of Wnt activity could be changed in different tissue types. However, this would involve additional training of the nearest centroid method to be applicable to other tissue types. One of the strengths of the Bayesian network model is that this tissue-specific training is not necessary, as it is established as being non-tissue type specific.EXAMPLE 3: TARGET GENE SELECTION
[0147] A transcription factor (TF) is a protein complex (that is, a combination of proteins linked together in a specific structure) or a protein that is capable of regulating the transcription of target genes by binding to sequences of specific DNA, thus controlling the transcription of genetic information from DNA to mRNA. The mRNA directly produced due to this action of the transcription complex is referred to in this document as the “direct target gene”. Activation of the pathway can also result in more secondary gene transcription, called “indirect target genes”. In the following, Bayesian network models (such as exemplary probabilistic models) comprising or consisting of direct target genes, such as direct links between pathway activity and mRNA level, are preferred, however, the distinction between direct and indirect target genes is not it is always evident. Here, a method for selecting direct target genes using a scoring function based on available literature data is presented. However, accidental selection of indirect target genes cannot be ruled out due to limited information and biological variations and uncertainties.
[0148] Path-specific mRNA target genes were selected from within the scientific literature using a ranking system in which scientific evidence for a specific target gene was given a ranking depending on the type of scientific experiments in which the evidence was accumulated . While some evidence is merely suggestive of being a target gene, such as an mRNA growing in a microarray of an embryo, in which the Hedgehog pathway is known to be active, other evidence may be very strong, such as a combination of a transcription factor binding site of the identified pathway and obtaining this site in a chromatin immunoprecipitation assay (ChIP) after stimulation of the specific pathway in the cell and increase in mRNA after specific pathway stimulation in a cell line.
[0149] Several types of experiments to find pathway-specific target genes can be identified in the scientific literature:
[0150] ChIP experiments, in which the direct binding of a pathway transcription factor to its binding site in the genome is shown. Example: Using chromatin immunoprecipitation (ChIP) technology, subsequently, putative functional TCF4 transcription factor binding sites in the DNA of colon cell lines with and without active Wnt pathway were identified, as a subset of binding sites recognized purely with base to nucleotide sequence. Putative functionality was identified as ChIP-derived evidence that the transcription factor was shown to bind to the DNA binding site.
[0151] Electrophoretic Mobility Shift Assays (EMSA), which show the in vitro binding of a transcription factor to a DNA fragment containing the binding sequence. Compared to the ChIP-based evidence, the EMSA-based evidence is less strong as it cannot be translated to the in vivo situation.
[0152] Pathway stimulation and measurement of mRNA profiles in a microarray or using RNA sequencing, using pathway-inducible cell lines and measurement of mRNA profiles measured at various points in time after induction — in the presence of cycloheximide, which inhibits protein translation, thus it is assumed that the induced mRNAs are direct target genes.
[0153] Similar to 3, but using quantitative PCR to measure the amounts of mRNAs.
[0154] Identification of transcription factor binding sites in the genome using a bioinformatics approach. Example for the Wnt pathway: Using the known transcription factor TCF4-beta catenin DNA binding sequence, a software program was run on the human genome sequence, and potential binding sites were identified, both in gene promoter regions and in other genomic regions.
[0155] Similar to 3, only in absence of cycloheximide.
[0156] Similar to 4, only in absence of cycloheximide.
[0157] mRNA expression profile of tissue samples or specific cells in which the pathway is known to be active, however, in the absence of the appropriate negative control condition.
[0158] In the simplest form, one can give every point of potential mRNA 1-target for each of these experimental approaches in which the mRNA target was identified.
[0159] Alternatively, points can be given incrementally, meaning that one technology, 1 point, second technology adds a second point, and so on. Using this relative ranking strategy, a list of more reliable target genes can be made.
[0160] Alternatively, classification otherwise can be used to identify target genes that are more likely to be direct target genes, giving a higher number of points to technology that provides more evidence for a direct target gene in vivo, in the above list, this would mean 8 points for experimental approach 1), 7 to 2) and going up to one point for experimental approach 8. Such a list could be called “general target gene list”.
[0161] Despite biological variations and uncertainties, the inventors assumed that direct target genes are more likely to be induced in a tissue-independent manner. A list of these target genes can be called an “evidence-cured list of target genes”. These cured target lists have been used to build computational models that can be applied to samples from different tissue sources.
[0162] The "general target gene list" likely contains genes that are more tissue-specific, and could potentially be used to optimize and increase the sensitivity of the model for application to specific tissue samples, such as breast cancer samples .
[0163] Next, it will be illustrated exemplarily how the selection of an evidence-cured list of target genes was specifically constructed for the ER pathway.
[0164] For the purpose of selecting ER target genes used as input to the "model", the following three criteria were used:
[0165] 1. The gene promoter/enhancer region contains an estrogen response element (ERE) motif:
[0166] a. It must be proven that the ERE motif responds to estrogen, for example, through a transient transfection assay where the specific ERE motif is linked to a reporter gene, and
[0167] b. The presence of the ERE motif must be confirmed, for example, by an enriched motif analysis of the gene promoter/enhancer region.
[0168] 2. ER (differentially) binds in vivo to the promoter/enhancer region of the gene in question, demonstrated, for example, by a ChIP/CHIP experiment or a chromatin immunoprecipitation assay:
[0169] a. It is proven that ER binds to the promoter/enhancer region of the gene when the ER pathway is active, and
[0170] b. (preferably) does not bind (or binds weakly) to the promoter/enhancer region of the gene if the ER pathway is not active.
[0171] 3. The gene is transcribed differently when an ER pathway is active, demonstrated, for example, by
[0172] a. enrichment in turn of the mRNA of the gene in question by real-time PCR, or microarray experiment, or
[0173] b. the demonstration that RNA Pol II binds to the promoter region of the gene through an immunoprecipitation assay.
[0174] Selection was made by defining as ER target genes those genes for which sufficient and well-documented experimental evidence was obtained proving that all three criteria mentioned above were met. A suitable experiment to collect evidence of differential binding to ER is to compare the results of, for example, ChIP/CHIP experiment on a cancer cell line that responds to estrogen (eg, the MCF-7 cell line), when exposed or not exposed to estrogen. The same goes for collecting evidence of mRNA transcription.
[0175] The above discusses the generic approach and a more specific example of the target gene selection procedure that was employed to select a range of target genes based on evidence found using the aforementioned approach. The lists of target genes used in the Bayesian network models for exemplary pathways, namely the Wnt, ER, Hedgehog and AR pathways, are shown in Table 1, Table 2, Table 3 and Table 4, respectively.
[0176] The ER pathway target genes used for the Bayesian network model of the ER pathway described in this document (shown in Table 2) contain a selection of target genes based on their evidence scores in the literature; only the target genes with the highest evidence scores (preferred target genes according to the invention) were added to this short list. The complete list of ER target genes, including also genes with a lower evidence score, is shown in Table 5.
[0177] A sub-selection or additional classification of target genes from the Wnt, ER, Hedgehog and Ar pathways shown in Table 1, Table 2, Table 3 and Table 4, was performed based on a combination of the literature evidence score and relationships calculated using the conditional probability tables trained linking the probeset nodes to the corresponding target gene nodes. The probability relationship is an assessment of the importance of the target gene in inferring pathway activity. In general, it is expected that the expression level of a target gene with a higher odds ratio is more likely to be more informative about the overall activity of the pathway compared to target genes with lower odds ratios. However, due to the complexity of cell signaling pathways, it must be understood that more complex interrelationships may exist between target genes and pathway activity - for example, consideration of expression levels of various combinations of target genes with low odds ratios may be more evidential than considering target genes with higher odds ratios alone. In the Wnt, Hedgehog and AR modeling reported in this document, it was found that the target genes shown in Table 6, Table 7, Table 8 and Table 9 are of a more evidential nature for predicting the activities of the Wnt, ER, Hedgehog pathway and AR compared to the lowest ranked target genes (thus the target genes shown in Tables 6 to 9 are particularly preferred in accordance with the present invention). However, considering the relative ease with which acquisition technologies such as microarrays can acquire expression levels for large sets of genes, the use of some or all of the target genes in Table 6, Table 7, Table 8 is contemplated. and Table 9, and optionally additionally using one, two, some or all of the additional target genes from classifications shown in Table 1, Table 2, Table 3 and Table 4, in the Bayesian model illustrated in Figure 6.
[0178] Table 1. Evidence-cured list of target genes of the Wnt pathway used in the Bayesian network and associated probe sets to measure the mRNA expression level of target genes (no. = sequence number in attached sequence listing) .




[0179] Table 2. Evidence-cured list of target genes of the ER pathway used in the Bayesian network and associated probe sets to measure the mRNA expression level of the target genes (no. = sequence number in the attached sequence listing ).




[0180] Table 3. Evidence-cured list of target genes of the Hedgehog pathway used in the Bayesian network and associated probe sets to measure the mRNA expression level of the target genes (no. = sequence number in the attached sequence listing ).




[0181] Table 4. Evidence-cured list of target genes of the AR pathway used in the Bayesian network and associated probe sets to measure the mRNA expression level of the target genes (no. = sequence number in the sequence listing attached).



[0182] Table 5. Gene symbols of the ER target genes that showed significant evidence in the literature (= long list of ER target genes) (n° = sequence number in the attached sequence listing).


[0183] Table 6. Short list of Wnt target genes based on evidence score in literature and probability relation (# = sequence number in attached sequence listing).



[0184] Table 7. Short list of ER target genes based on literature evidence score and probability relationship (n° = sequence number in attached sequence listing).


[0185] Table 8. Short list of Hedgehog target genes based on literature evidence score and odds ratio (# = sequence number in attached sequence listing).


[0186] Table 9. Short list of RA target genes based on literature evidence score and probability relationship (# = sequence number in attached sequence listing).


EXAMPLE 4: COMPARISON BETWEEN A LIST CURED BY EVIDENCE AND A LIST OF BROAD LITERATURE
[0187] The list of Wnt target genes constructed based on evidence from the literature following the procedure described in this document (Table 1) is compared with another list of target genes that does not follow the procedure mentioned above. The alternative list is a compilation of genes indicated by a variety of data from various experimental approaches to be a Wnt target gene published in three public sources by renowned laboratories, known for their experience in the field of molecular biology and with the Wnt pathway. The alternative list is a combination of the genes mentioned in Table S3 by Hatzis et al. (Hatzis P, 2008), the text and table S1A by de Sousa and Melo (de Sousa E Melo F, 2011) and the list of target genes collected and maintained by Roel Nusse, a pioneer in the field of Wnt signaling ( Nusse, 2012). Combining these three sources resulted in a list of 124 genes (= list from the broad literature, see Table 10). Here, the issue of performance in predicting the activity of Wntem clinical samples by the algorithm derived from this alternative list having a similar or better performance compared to a model built on the existing list of genes (= evidence-cured list, Table 1) is discussed.
[0188] Table 10. Alternative list of Wnt target genes (= list from extensive literature) (n° = sequence number in attached sequence listing).






[0189] The next step consisted of discovering the Affymetrix® GeneChip Human Genome U133 Plus 2.0 array probe sets that matched the genes. This process was carried out using the Bioconductor plugin in R and manual curation for the relevance of probesets based on the UCSC Genome Browser, thereby removing, for example, probesets on opposite strands or exon regions of external genes. For two of the 124 genes, there are no probesets available on this microarray chip and therefore could not be inserted into the Bayesian network, these are LOC283859 and WNT3A. In total, 287 sets of probes were shown to match the remaining 122 genes (Table 11).
[0190] Table 11. Probe sets associated with the Wnt target genes in the gene list in the extensive literature (n° = sequence number in the attached sequence listing).





[0191] Subsequently, the Bayesian network was built similar to Figure 6 and the procedure explained in this document. Analogously to the description of the Wnt Bayesian network based on the evidence-cured list, the conditional probability tables of the edges between sets of probes and their respective genes, both the evidence-cured list and the list from the broad literature, were trained using data processed by fRMA from 32 normal colon specimens and 32 adenoma specimens from the Gene Expression Omnibus dataset GSE8671 (accessible at http://www.ncbi.nlm.nih.gov/geo/, last accessed July 13, 2011) .
[0192] The trained Bayesian networks were then tested on several datasets to infer the probability P(Wnt On) that the Wnt pathway is "on", that is, active, which is taken equal to the inferred probability of that the transcription complex of the Wnt pathway is “present”. The summarized results of the trained literature-wide model and the evidence-cured model are shown in Figures 14 to 19.
[0193] Of course, it could be deduced that the model in the broad literature generally predicts more extreme probabilities for Wnt signaling to be on or off. In addition, the alternative model predicts similar results for the colon cancer datasets (GSE20916, GSE4183, GSE15960) but, more than expected, samples with the predicted Wnt signaling in breast cancer sample datasets (GSE12777), liver cancer (GSE9843) and medulloblastoma (GSE10327).
[0194] In conclusion, the list of target genes from the extensive literature results in roughly equally good predictions of Wnt activity in colon cancer on the one hand, but worse predictions (many false positives) in other cancers on the other hand. This may be a result of the alternative list of target genes being too skewed towards colon cells specifically, thus too tissue-specific; the main interest of both Sousa E Melo et al. as well as Hatzis et al. was colorectal cancer, although non-colon-specific Wnt target genes may be included. In addition, non-Wnt-specific target genes possibly included in these lists may be a source of worsened predictions of Wnt activity in other cancers. The alternative list probably contains more indirectly regulated target genes, which probably makes it more tissue-specific. The original list is adjusted to contain direct target genes, which are more likely to represent genes that are sensitive to Wnt in all tissues, thus reducing tissue specificity.EXAMPLE 5: BAYESIAN NETWORK TRAINING AND USE
[0195] Before the Bayesian network can be used to infer the activity of roads in a sample of tests, the parameters describing the probabilistic relationships between the network elements have to be determined. Furthermore, in the case of discrete states of the input measurements, it is necessary to establish thresholds that describe how to perform the discretization.
[0196] Typically, Bayesian networks are trained using a representative set of training samples, of which preferably all states of all network nodes are known. However, it is not very practical to obtain training samples from many different types of cancer, of which the activation state of the pathway to be modeled is known. As a result, the available training sets consist of a limited number of samples, typically only from one type of cancer. To allow the Bayesian network to generalize well to other sample types, therefore, special attention must be paid to the way in which the parameters are determined, which is preferably done as follows, in the approach described in this document.
[0197] For the TF node, the (unconditional) probability of being in “absent” and “present” state is given by the expected occurrence in a large set of samples. Alternatively, they can be set to 0.5, as is done in Figure 1, so there is no bias towards a positive or negative result.
[0198] For the target gene nodes, the conditional probabilities are set as in Figure 1. If the TF element is “missing”, it is more likely that the target gene is “low”, thus a probability of 0.95 is chosen for this, and a probability of 0.05 for the target gene to be “high”. This last (non-zero) probability is to take into account the (rare) possibility that the target gene is regulated by other factors or accidentally observed “high” (eg due to measurement noise). If the TF element is “present”, then with a fair probability of 0.70, the gene is “high”, and with a probability of 0.30, the target gene is “low”. These latter values are chosen in this way because there may be several reasons why a target gene is not highly expressed even though the TF element is present, for example because the promoter region of the gene is methylated. In case a target gene is not up-regulated by the TF element, but down-regulated, the probabilities are chosen in a similar way, but reflecting down-regulation with the presence of the TF element.
[0199] For the Bayesian network model given in Figure 6, where the intensities of the probe sets form the input measurements, one must finally determine the parameters for discretization and for the conditional probability tables relating the intensities of the probe sets the mRNA levels of the respective target genes. Both are based on training data on the current invention. For the discretization of the intensity level of a set of probes in “low” and “high” states, an adequate threshold is determined, which better separates the intensity values in a set of training samples, where the pathway is activated ( “on” samples) from the intensity values in a set of training samples in which it is not (“off” samples). Finally, conditional probability tables describing the probabilities of a set of probes having a “low” or “high” intensity depending on the “low” or “high” state of the respective target gene are made by counting the number of “on” samples and “off” with a probe set intensity value below and above the respective threshold. This is known in the literature as the frequentist approach. A dummy count is added to each group to avoid entries in the conditional probability tables with a value of zero, to avoid extreme Bayesian network behavior.
[0200] After the Bayesian network has been trained, it can be applied to a test sample as follows, considering the Bayesian network in Figure 6, and assuming that microarray measurements related to the probe sets are available. The first step is to discretize the input measurements by comparing the intensity of each set of probes in the test sample to the respective threshold as described above. This comparison can be made concretely, setting each set of probes to "low" or "high" intensity (called 'concrete evidence'), it can be done smoothly, assuming a certain uncertainty (noise) in the measurement, setting for each set of probes a probability of being “low” or “high” (called 'soft evidence'). For example, soft evidence from a set of probes with an intensity just below the threshold might be a 0.8 probability of being “low” and a 0.2 probability of being “high”, based on an adequate estimate of noise and the difference to the threshold.
[0201] Next, this concrete or soft evidence is provided to an inference mechanism suitable for Bayesian networks, for example, based on a join tree algorithm (see (Neapolitan, 2004)). Such a mechanism can then infer the updated probability of the TF element being “absent” or “present”, considering the evidence provided. The inferred probability of the TF element being “present” is then interpreted as the estimated probability that the respective pathway is active.
[0202] Preferably, the training of the Bayesian network models of the Wnt, ER, Hedgehog and AR roads is done using public data available in the Gene Expression Omnibus (accessible at http://www.ncbi.nlm.nih.gov/geo/, cf . above).
[0203] The Bayesian network was trained exemplarily using 32 colon samples considered to have an inactive Wnt pathway and 32 confirmed adenoma samples known to have an active Wnt pathway (data set from GSE8671).
[0204] The Bayesian network of the ER pathway was exemplary trained using 4 estrogen-deprived MCF7 samples, known to have an ER pathway, and 4 estrogen-stimulated MCF7 samples, considered to have an ER pathway, from the set of GSE8597 data also accessible in Gene expression Omnibus.
[0205] The Bayesian network model of the Hedgehog pathway was exemplary trained using 15 basal cell carcinoma samples confirmed to have an active Hedgehog pathway and 4 normal skin cell samples representing samples with an inactive Hedgehog pathway available in the GSE7553 dataset.
[0206] The Bayesian network model of the AR pathway was trained exemplarily using 3 samples with positive AR activity, LNCaP cell lines stimulated with Dihydrotestosterone (DHT), a potent AR pathway activator, and 3 unstimulated LNCaP cell lines representing the case of the inactive RA pathway.
[0207] Referring to Figure 35 and Figure 36, Bayesian network models of the Wnt and ER pathways were used to predict pathway activities in similar samples (Colon and MCF7 breast cancer cell line samples for Bayesian network of Wnt and Er, respectively) not used in the training procedure as described in this document (no dataset suitable for the Bayesian Hedgehog network was found). The expected route activities of the vast majority of samples should be in line with the clinically expected route activities for the model to be validated.
[0208] Figure 35 shows the predicted Wnt activities, illustrated as the logit of P(Wnt on) on the vertical axis, for the samples, illustrated by the bars on the horizontal axis, of the colon samples grouped by classification, indicated by the color of the bars, in the dataset of GSE20916. All normal colon samples are legitimately predicted to have an inactive pathway (score < 0) based on the fact that this is a healthy tissue sample. All but four samples considered to have an active pathway are predicted to have an active Wnt pathway.
[0209] In Figure 36, the validation results of the trained ER Bayesian network model are shown for two microarrays measured using a MCF7 breast cancer cell line sample, one stimulated with estrogen (E2) and the other with a control negative (EtOH), originating from the GSE9253 dataset. In agreement with the claimed ER activity, the estrogen-stimulated sample is predicted to have an active ER pathway, whereas the negative control predicts an inactive ER pathway.
[0210] Additional details and examples for using trained Bayesian networks (eg from the Wnt, ER, AR and Hedgehog pathways) to predict the respective pathway activities are explained in Example 6 below.
[0211] The training process mentioned above can be used in other Bayesian networks for clinical applications. Here, it is shown and proven to work for Bayesian network models constructed using the method disclosed in this document representing cell signaling pathways, more specifically the Wnt, ER, AR and Hedgehog pathways.EXAMPLE 6: PATHWAY ACTIVITY DIAGNOSIS(ABNORMAL)
[0212] The following will illustrate exemplarily how to use, for example, Bayesian network models to diagnose the activity of a cell signaling pathway.
[0213] Bayesian networks of the Wnt, ER, Hedgehog and AR pathways, constructed using a node for the presence of transcription factor, a layer of nodes representing the mRNA of the target genes and a layer of nodes representing the intensities of the sets of probes corresponding to the target genes (Table 1, Table 2, Table 3 and Table 4), analogous to Figure 6 described in this document, and trained as described in this document, were used to predict pathway activity as "on", i.e. , active, or “off”, that is, inactive, on various datasets not previously used for training, to infer how well the inference component operates. Pathway activity scores are correlated with clinical knowledge. Results summaries for a selection of test runs are shown in Figures 21 seq.
[0214] Referring to Figures 21 seq., the results of inference of the pathway activity for medical tissue samples using the Bayesian network model described in this document are shown.
[0215] Figure 21 shows results for tests of Wnt activity in the GSE4183 colon sample dataset. The Bayesian network model yielded high P(Wnt On) values for the adenoma samples, and low values for the normal samples, which correspond to the (path)physiology of adenoma and healthy tissue. Healthy tissue has a slow cell proliferation and thus a low Wnt activity relative to adenomatous tissue, which has a rapid cell proliferation and thus high Wnt activity. For the IBD samples, the Bayesian network model showed low Wnt pathway activity (P(Wnt On)~0) for all samples but one. Again, this is consistent with the IBD samples not undergoing rapid cell proliferation. For colorectal cancer cell samples, results were mixed, with high Wnt pathway activity being detected in about half of these samples, but this may be a result of other pathways taking on the role of tumor booster when benign adenomatous tissue becomes malignant cancer tissue, or sample analysis problems, for example, the sample contains too much non-tumor tissue, or the mRNA is partially degraded
[0216] The Bayesian network model used in the experiments reported in this document was trained using the GSE8671 sample dataset. However, the Wnt pathway is present (though possibly inactive) in other cell types. It was therefore considered possible that the Bayesian network could be applicable to infer abnormally high Wnt pathway activity correlating with other types of cancers. The rationale for this is that, although the Bayesian network model was trained using colon samples, it was based on first principles of operation of the Wnt pathway present (though possibly inactive) in other cell types. Figures 22 to 24 show some results investigating such “multi-tissue type” inferences.
[0217] Figure 22 shows results for tests using the trained Bayesian network model using colon samples being applied to infer Wnt pathway activity in medulloblastoma samples (dataset GSE10327). The samples included in this dataset were further characterized in several subsets, one of them being samples with the Wnt pathway being active. The Bayesian network of the Wnt pathway predicts the active Wnt sample group having an active Wnt pathway, while the other samples were correctly predicted having an inactive Wnt pathway.
[0218] The results of tests using the Wnt Bayesian network model on a dataset containing liver cancer samples (GSE9843) are shown in Figure 23. Here, the samples are grouped by the following a priori annotations assigned by the set of GSE9843 data: “CTNNB1”, “Inflammation”, “Polysomy chr7”, “Proliferation”, and “Not noted”. Samples from the “Inflammation” group are uniformly inferred to not have abnormally high Wnt pathway activity, as expected, since the condition of inflammation does not imply rapid cell proliferation. Samples labeled “Polysomy chr7” are uniformly inferred to have no abnormally high Wnt pathway activity. The polysomy of chromosome number 7 means that there are more than two chromosome number 7. Since there is no reason to expect this polysomy condition to impact the Wnt pathway, it is not unexpected that these samples do not have abnormally high Wnt pathway activity.
[0219] About one in five of the samples labeled “Proliferation” had P(Wnt On)>0.5. Proliferation suggests a state of rapid cell multiplication. This state may be associated with abnormally high Wnta pathway activity, but it may also be associated with several other possible causes of cell proliferation. Therefore, about one in five of these samples having abnormally high Wnt pathway activity is not an absurd result.
[0220] About half of the samples from the “CTNNB1” group are inferred by the Bayesian network to have abnormally high Wnt pathway activity. The CTNNB1 gene encodes the beta-catenin protein, which is a regulatory protein of the Wnt pathway, and activation of mutations in this gene causes abnormal Wnt activation. Thus, a correlation between the “CTNNB1” group and high Wnt pathway activity is in line with expectations.
[0221] Figure 24 illustrates the test results of the Wnt Bayesian network model described in this document for a set of breast cancer samples. In this case, three groups of breast cancer cell lines are tested: a group for which the Wnt pathway is a priori known to be operating at an abnormally high level (group Wnt turned on); a group for which the Wnt pathway is a priori known not to be operating at an abnormally high level (group Wnt off); and another group for which the Wnt pathway activity is not known a priori (Unknown group); in addition, there is also a sample that is suspected of having a low level of Wnt activation (suspected Wnt), although there is a conflicting report in the literature that it may have an active Wnt pathway (but this is a minority report; more articles report an inactive Wnt pathway). As seen in Figure 24, the correlation of inferences provided by the Bayesian network with a priori knowledge is strong for the Wnt on and Wnt off groups. In addition, the sample on the right-hand side of the graph (suspected Wnt) presents an inference that corresponds to the majority of reports in the literature claiming that the Wnt pathway is turned off. In the case of the unknown group shown in Figure 24, for which there is no prior knowledge of the Wnt pathway activity, the Bayesian network infers low activity for the Wnt pathway, except for one case for which P(Wnt On)>0, 5; the literature shows that this cell line has a high expression of the LRP6 co-receptor, which may explain that the Wnt pathway is linked.
[0222] Figure 25 shows the results for the same dataset from breast cancer cell lines, but now tested for ER activity using the Bayesian ER network trained using MCF7 breast cancer cell lines as described in this document . Samples known a priori to have an active Wnt pathway were predicted to have an inactive ER pathway, which is not surprising since the Wnt pathway is already driving rapid cell multiplication. ER positive samples, on the other hand, are found between the Wnt off samples and the unknown samples. In view of Figure 24, this is not surprising.
[0223] Test results of the Bayesian network predictions of ER trained in breast cancer cell lines for a set of cancer samples (GSE12276) are shown in Figure 26. The breast cancer samples have been subdivided into known classifications: Luminal A (LumA), Luminal B (LumB), Human Epidermal Growth Factor Receptor 2 positive (HER2), and basal breast cancer subtype. Tissue samples in the luminal A and luminal B subtypes are known to express ER. It is also in these subtypes that most samples are predicted to have high ER pathway activity. On the other hand, samples that are classified as being of the basal subtype are known to have no ER expression or low expression, which correlates well with no active ER pathway predicted in the basal group samples. In the HER2 group, only three samples have a P(ER bound) >0.5, whereas most samples are predicted to have an inactive ER pathway. This correlates well with the fact that the classification is done on the fact that these samples have amplified HER2 expression; uncontrolled cell replication is presumably triggered through HER2 signaling via cell signaling pathways other than the ER pathway (see, for example, Wnt's active breast cancer cell lines in Figure or cancer samples from active breasts of Hedgehog in Figure 30).
[0224] The ER Bayesian network model built and trained as described in this document is used to predict ER pathway activity in a large panel of cell lines from various cancers, the results are shown in Figure 27. As expected, only samples predicted to be in active ER were found in breast cancer cell lines. All other types of cancer cell lines were predicted to have an inactive ER pathway, which is as expected.
[0225] The Bayesian network model built and trained for the Hedgehog pathway as described in this document is used to predict the activity of the Hedgehog pathway for cell lines of different types of cancer in the GSE34211 dataset. The predictions of Hedgehog activity are shown in Figure 28. The largest fractions of Hedgehog activity predicted to be positive are found in the central nervous system (CNS), skin, endometrium, and uterus cancers, which is in good agreement with the knowledge of the literature regarding Hedgehog-dependent cell proliferation in these cell types.
[0226] Figure 29 shows the predicted Hedgehog activity of medulloblastoma samples (GSE10327) that has already been analyzed using the Wnt Bayesian network model as described in this document. The medulloblastoma samples have been characterized into subclasses, with one of them having an active Hedgehog signaling pathway (identifier: SHH). All samples in the SHH subtype are predicted to have active Hedgehog signaling. Furthermore, medulloblastoma samples in the Wnt subtype were also predicted to have an active Hedgehog pathway. This is in agreement with clinical evidence showing that both pathways are often active in these tumors. However, the Bayesian network of Wnt was clearly able to correctly predict Wnt activity only in the Wnt subtype. Thus, the combination of the Bayesian network of Wnt and Hedgehog is able to make a correct classification of these two subtypes.
[0227] The predicted Hedgehog activity in the GSE12276 breast cancer samples, previously used to predict the ER activity using the Bayesian network model of ER, using the Bayesian network model of Hedgehog is shown in Figure 30. The Hedgehog pathway is predicted to be active in a fraction of samples of each subtype. This seems odd, but in correspondence with the prediction of the ER pathway shown in Figure 26, it can be seen that Hedgehog activity is only predicted in samples that do not have an active ER pathway. This is in good agreement with the hypothesis that uncontrolled cell proliferation in (breast) tissue may be triggered by different signaling pathways.
[0228] In summary, the test results for several tissue and cancer cell samples presented in Figures 21 to 30 strongly suggest that the Bayesian networks of the Wnt, ER, and Hedgehog models trained on tissue/lane specific samples are applicable to sample analysis of other types of fabric. This can allow cell signaling pathway analysis to be applied to “multi-types of tissues”. Thus, the CDS 10 system (as described in this document) is readily applied to assess pathway activity in a range of tissue types other than the tissue type of the samples used to train the Bayesian Network 40 model (see, for example, Figure 20, which schematically shows a clinical decision support system (CDS) configured to assess one or more cell signaling pathways as disclosed in this document (example shown for the Wnt pathway)). In cases where inference components 40, 44, 46, 48 indicate that the tissue under analysis exhibits abnormally high Wnt, ER or Hedgehog pathway activity, but no tissue-specific drug is available, a Wnt, ER, or pathway suppressing drug General hedgehog, or a drug specific to the malfunction, may be considered by the clinician based on recommendation 28 or recommendation 26, respectively, as provided by the CDS 10 system.
[0229] Although the results of Figures 21 to 30 indicate applicability in several tissue types of the Bayesian network model for the Wnt, ER and Hedgehog pathway, it is expected that, for clinical applications, the Bayesian network models can optionally be updated or tailored to maximize their applicability to the specific tissue type under consideration (eg, breast tissue or liver tissue). Such an update or adaptation could, for example, imply the adjustment of conditional probabilities based on clinical studies of the type of tissue under analysis or the enrichment of the list of evidence-cured target genes, described in this document, with tissue-specific target gene of the route(s) under investigation. In addition, nodes can be added or removed to better fit the Bayesian network model to the tissue under analysis. Alternatively, different Bayesian network models can be trained from scratch using different training sets for different tissue types. In addition, the results in Figures 21 to 30 illustrate the ability of the process described in this document to develop and train Bayesian network models using target gene lists curated by evidence from pathways other than Wnt, ER, and Hedgehog, to predict and diagnose road activity.
[0230] Test results from the Bayesian network model of AR constructed and trained as described in this document were exemplary used to predict AR activity in LNCaP prostate cancer cell lines treated with different treatment regimens (GSE7708) (see Figure 38). As expected, LNCaP cells not stimulated with DHT result in a predicted inactive AR pathway, whereas LNCaP stimulated cells were correctly predicted to have an active AR pathway and LNCaP cells treated with Bicalutamide, an anti-androgen drug, to have an inhibited AR pathway.
[0231] The trained AR pathway Bayesian network as described in this document was also used to predict the likelihood of the RA pathway being active in prostate cancer samples from the GSE17951 dataset (the results are shown in Figure 39). Most prostate and tumor biopsies were, not unexpectedly, predicted to have a higher probability of AR activity compared to control samples.
[0232] The Bayesian network model of RA was also applied to a test of multiple tissues, namely the breast cancer samples included in the GSE12276 dataset. The results for this test are shown in Figure 40. A small fraction of the samples, found in each subgroup, are predicted to have an active pathway, whereas the vast majority of samples had an inactive RA pathway. Interestingly, the highest percentage of samples with an active AR pathway are found in the HER2 subgroup, which is not unexpected, as it is known in the literature that there is interference between the HER2 and AR pathways and the AR pathway can also be induced by HER2 signaling.
[0233] The Bayesian AR network model mentioned above was also used to predict the AR pathway activity in two sets of cell line samples from different types of cancer (GSE36133 and GSE34211) as illustrated in Figure 41 and Figure 42. As expected, most cell lines have been shown to have an inactive AR pathway. Exceptions to this are prostate cancer samples with multiple cancer cell line samples expressing AR pathway activity. In Table 12, it is shown that all predictions of AR pathway activity from the prostate cancer samples are in agreement with the known AR activity.
[0234] Table 12. Known and predicted AR activity in prostate cancer cell lines in the GSE36133 and GSE34211 datasets.

EXAMPLE 7: PROGNOSIS BASED ON ROAD ACTIVITY
[0235] Early developmental pathways such as Wnt and Hedgehog are thought to play a role in metastasis caused by cancer cells that have reverted to a more stem cell-like phenotype called cancer stem cells. Indeed, sufficient evidence is available that early developmental pathways, such as the Wnt pathway, play a role in cancer metastasis, allowing metastatic cancer cells to share seed allocation in another tissue or organ. Metastasis is associated with poor prognosis, therefore, the activity of early developmental pathways, such as the Wnt and Hedgehog pathways, in cancer cells is expected to be predictive of poor prognosis. This is confirmed by the fact that breast cancer patients from the GSE12276 dataset who were identified as having an active ER pathway but not having an active Wnt or Hedgehog pathway using the Bayesian network models described in this document had a better prognosis than patients identified as having a Hedgehog or Wnt pathway or both, as illustrated by the Kaplan-Meier curve in Figure 37. EXAMPLE 8: THERAPY PLANNING, DRUG EFFECTIVENESS FORECAST, SIDE EFFECTS FORECAST AND DRUG EFFECTIVENESS MONITORING
[0236] The following example illustrates how to use probabilistic models, in particular Bayesian network models, for therapy planning, drug efficacy prediction, drug efficacy monitoring and related activities.
[0237] The Bayesian network model of the ER pathway, built using a node for the presence of transcription factor, a layer of nodes representing the mRNA levels of the target genes (Table 2) and a layer of nodes representing the intensities of the probe sets corresponding to the target genes (Table 2), analogous to Figure 6 described in this document, and trained as described in this document, were used to predict pathway activity. Pathway activity is further shown to be correlated with drug efficacy or drug efficacy monitoring. Results summaries are shown in Figures 31 and 32.
[0238] Tamoxifen is a drug currently used for the treatment of ER+ (estrogen receptor positive) breast cancer. It acts as an estrogen receptor antagonist, inhibiting uncontrolled cell proliferation, which is believed to be induced by ER signaling. Unfortunately, not all breast cancer responds to treatment with Tamoxifen, despite the demonstration of the presence of ER protein in cancer cells by routine histopathology analysis of cancer tissue slides. Many studies have been conducted to investigate this so-called resistance to tamoxifen. The publicly available GSE21618 dataset is the result of one such study and contains microarray data from tamoxifen and wild-type MCF7 resistant cell lines under different treatment regimens. The ER Bayesian network model constructed and trained as described in this document is used to analyze Tamoxifen and MCF7 resistant cell lines under different treatment regimens, the results are shown in Figure 31.
[0239] The control Tamoxifen-resistant cell line, indicated by TamR.Ctrl, is predicted to have an inactive ER pathway for every time point after addition of Tamoxifen (1, 2, 3, 6, 12, 24 and 48 h ). It is not surprising that the treatment of the Tamoxifen resistant cell line, which is insensitive to the Tamoxifen treatment indicated by TamR.Tam, is ineffective, which is also illustrated by the predicted inactivity of the ER pathway for this group over the same time points . According to the Tamoxifen-resistant cell line analysis (TamR.Ctrl), the driving force of uncontrolled cell proliferation is not due to active ER signaling; therefore, treating it with an ER antagonist will not inhibit cell proliferation. This illustrates that Tamoxifen treatment is not recommended in case of predicted negative ER pathway activity.
[0240] On the other hand, the wild-type MCF7 cell line, known to be sensitive to Tamoxifen, treated with 17beta-estradiol (wt1.E2) slowly reacts to hormonal treatment, which is visible in the increasing predictions of positive ER activity. Treatment of such a cell line with aromatase inhibitors that are known to inhibit estrogen production will inhibit the ER pathway, which is illustrated by the time-decreasing ER pathway prediction. Supporting this are the ER pathway predictions made based on microarray data from MCF7 samples with estrogen to increase time in the GSE11324 dataset, the results shown in Figure 32.
[0241] The above illustrates the ability of probabilistic models, in particular, Bayesian network models, to be used for therapy planning, drug efficacy prediction, and drug efficacy monitoring. However, it should be understood that the same methodology would also apply to the prediction and monitoring of adverse effects.EXAMPLE 9: DRUG DEVELOPMENT
[0242] Similar to monitoring response to therapy, a pathway model can be used in drug development to assess the efficacy of various putative compounds. For example, when screening many compounds for a possible effect on a certain pathway in a cancer cell line, the respective pathway model can be used to determine whether pathway activity goes up or down after application of the compound or not. Often this check is done using only one or a few putative markers of pathway activity, which increases the chance of ineffective monitoring of the treatment effect. Furthermore, in follow-up studies in animal subjects or patients, pathway models can be used in the same way to assess the efficacy of drug candidates, and determine an optimal dose to have maximum impact on pathway activity.
[0243] An example of ineffective monitoring of new drug compounds is illustrated by the predicted AR pathway activity in the GSE7708 samples as shown in Figure 38. In this study, two possible drug compounds to inhibit the AR pathway activity, denoted by Polyamide 1 and Polyamide 2 have been developed. These two polyamides were shown to be capable of inhibiting the AR pathway based on the findings that polyamides bind to the Androgen Response Element (ARE) and inhibit the expression of KLK3 (=PSA), a well-known marker for AR activity as well. included in the selection of target genes as described in this document, as well as « 35% of the transcripts that were induced by DHT. In contrast, the Bayesian network model of the AR pathway predicted that these samples still had an active AR pathway. Investigation of the inferred probabilities of target genes being up-regulated using the Bayesian AR network model indicated that KLK3, in contrast to the other target genes, was down-regulated according to the findings, while all other target genes (except AR, GUGY1A3 and TMPRSS2 in the case of Polyamide 1) were expressed differently in samples treated with Polyamide 1 and Polyamide 2. In other words, only one marker for AR activity, KLK3, was down-regulated, whereas most of the The identified target genes were still up-regulated, indicating that the Ar pathway is still largely intact, and thus, active. By taking into account a larger number of target genes based on evidence from the literature, the inventors were able to show that the inhibition of the AR activity of polyamides is limited and that only KLK3 expression is clearly down-regulated using these polyamides. Furthermore, this illustrates the value of a systematic approach using a Bayesian network model compared to a reductionist approach in drug development.EXAMPLE 10: ASSAY DEVELOPMENT
[0244] Instead of applying the aforementioned Bayesian networks to mRNA input data from microarrays or RNA sequencing, it may be beneficial in clinical applications to develop dedicated assays to perform sample measurements, for example, on an integrated platform, using qPCR to determine mRNA levels of target genes. The RNA/DNA sequences of the revealed target genes can then be used to determine which primers and probes to select on such a platform.
[0245] The validation of such a dedicated assay can be done using microarray-based Bayesian networks as a reference model, and verifying if the developed assay provides similar results in a set of validation samples. Alongside a dedicated assay, this can also be done to build and calibrate Bayesian network models using mRNA sequencing data as input measurements.EXAMPLE 11: ROUTE SURVEY AND CANCER PATHOPHYSIOLOGY SURVEY
[0246] The following will illustrate how Bayesian network models can be employed in pathways (clinical) surveys, that is, surveys interested in finding out which pathways are involved in certain diseases, which can be followed up to more detailed surveys, eg for link mutations in signaling proteins to changes in pathway activation (measured with the template). This is relevant for investigating the initiation, growth and evolution and metastasis of specific cancers (the pathophysiology).
[0247] The Bayesian network models of the Wnt, ER, Hedgehog and AR pathways, constructed using a node for the presence of transcription factor, a layer of nodes representing the mRNA levels of target genes (Table 1, Table 2, Table 3 and Table 4) and a layer of nodes representing the probe sets intensities corresponding to the target genes (Table 1, Table 2, Table 3 and Table 4), similar to Figure 6 described in this document, and trained as described in this document, were used to predict pathway activity from a dataset consisting of breast cancer samples (GSE12276).
[0248] Suppose the researcher is interested in analyzing the cell signaling pathway or pathways and the specific dysregulation(s) that trigger(s) uncontrolled cell proliferation. The researcher can analyze the microarray data using the mentioned probabilistic models, in particular the Bayesian network models, to discover which pathways are presumably the cause of uncontrolled cell proliferation. Shown in Figure 33 and Figure 34, you can see an illustration of such an analysis for the case of Wnt, ER, and Hedgehog activity (basal and luminal A samples from dataset GSE12276). Subsequently, the researcher can search in greater detail to find the exact cause of road dysregulation.
[0249] Referring to Figure 34, basal samples are known to have triple negative receptor status (ER, PR and HER2), so it is not surprising to see that all samples are predicted to have an inactive ER pathway. On the other hand, some of the samples are predicted to have Wnt or Hedgehog, or both, active, as shown in Figure 34. These predicted pathway activities persuade the researcher to investigate these samples in greater detail for, for example, known mutations or other known dysregulations in the Wnt and/or Hedgehog pathways.
[0250] Another example is given in Figure 33, where the activities of Wnt, ER and Hedgehog in luminal A samples from the GSE12276 dataset are illustrated. Deluminal A samples are known to express ER, however, this does not necessarily mean that the cancerous properties are due to active ER signaling. From the predicted pathway activities, it can be inferred that less than half of the samples have an active ER signaling. However, some of the samples that do not have active ER signaling are shown to have an active Wnt and/or Hedgehog pathway. This may allow the researcher to investigate these samples in greater detail for defects in the Wnt and/or Hedgehog signaling pathway, respectively. Some of these samples do not predict any of the three included pathways being active; perhaps other pathways are causing the uncontrolled cell proliferations. In addition, this gives the researcher additional information to look for defects in other avenues.
[0251] In summary, the illustrations described in this document indicate the ability of trained Bayesian network models (as described above) to confirm the process of discovering the cause of uncontrolled cell proliferation in a more targeted method. By employing Bayesian networks to screen samples for pathway activities, predicted pathway activities can identify possible pathways for cell proliferation, which can be followed up for more detailed research, for example, to link mutations in signaling proteins or other dysregulations to changes in activation (as measured by model).
[0252] As described in this document, the process to develop and train a Bayesian network of cell signaling pathways can be used to build a Bayesian network model for other pathways that could also be employed in connection with the present invention. INDIVIDUAL REGISTRATION IN A CLINICAL TRIAL BASED ON THE PLANNED ACTIVITY
[0253] If a drug candidate is developed to, for example, block the activity of a verta pathway that activates tumor growth, and this drug is entering the clinical trial, then an appropriate selection of individuals to enroll in such a trial is essential to prove the potential efficacy of the drug. In such a case, patients who do not have the respective pathway activated in their tumors should be excluded from the trial, as it is obvious that the drug cannot be effective if the pathway is not activated in the first place. Thus, a pathway model that can predict pathway activity can be used as a selection tool, to only select patients who are predicted to have the respective pathway activated.EXAMPLE 13: SELECTION OF POSTERIOR TEST(S) TO BE CARRIED OUT
[0254] If a tumor is analyzed using different pathway models, and the models predict a certain pathway dysregulation, then this may guide the selection of subsequent tests to be performed. For example, a proximity binding assay (PLA) can be performed to confirm the presence of the respective transcription complex (Soderberg O, 2006). Such a PLA can be designed to give a positive result if two key proteins in a TF complex actually bound, eg beta-catenin and TCF4 in the TF complex of the Wnt pathway.
[0255] Another example is that the predicted pathway to be deregulated is analyzed in greater detail in relation to the signaling cascade. For example, you can analyze key proteins in this pathway to determine if there are mutations in the regions of DNA encoding their respective genes, or you can test for the abundance of these proteins to see if they are taller or shorter than normal. . Such tests can indicate what the root cause behind the pathway's dysregulation is, and provide insights into which available drugs could be used to reduce pathway activity.
[0256] These tests are selected to confirm the pathway activity as identified using the Bayesian model. However, selection of associated diagnostic tests is also possible. After identifying the pathway using the template, to choose the targeted therapy, only the associated diagnostic tests need to be performed (the selection), which are applicable to the identified pathway.EXAMPLE 14: SELECTION OF ASSOCIATED DIAGNOSTIC TESTS
[0257] Similar to the previous example, if a tumor is analyzed and pathway models predict dysregulation of a certain pathway, and optionally, a series of additional tests have been performed to investigate the cause of dysregulation, then an oncologist can select a range of drugs candidates to treat the patient. However, treatment with such a drug may require that an associated diagnostic test be performed first, for example, to comply with clinical guidelines or ensure reimbursement of treatment costs, or because regulations (FDA) require the associated diagnostic test to be performed. before giving the drug. An example of such an associated diagnostic test is the Her2 test for treating breast cancer patients with the drug Herceptin (Trastuzumab). Thus, the result of the pathway models can be used to select the drug candidates and their associated diagnostic tests to be performed.EXAMPLE 15: CDS APPLICATION
[0258] Referring to Figure 20 (schematically showing a clinical decision support system (CDS) configured to assess one or more cell signaling pathways as disclosed in this document (example shown for the Wnt pathway)), a support system Clinical Decisions (CDS) 10 is implemented as a properly configured computer 12. Computer 12 may be configured to operate as CDS system 10 by running appropriate software, firmware or other instructions stored on a non-transient storage medium (not shown) such such as a hard disk or other magnetic storage medium, an optical disk or other optical storage medium, random access memory (RAM), read-only memory (ROM), flash memory, or other electronic storage medium, a network server, and so on. Although the illustrative CDS system 10 is realized by the illustrative computer 12, more generally, the CDS system may be realized by a digital processing device or an apparatus comprising a digital processor configured to perform clinical decision support methods as set forth. in this document. For example, the digital processing device can be a portable device (eg a personal data assistant or smartphone running a CDS application), a notebook computer, a desktop computer, a tablet computer or device, a network server remote, and so on. Computer 12 or other digital processing device typically includes or is operatively connected to a display device 14 through which information, including clinical support recommendations, is displayed to medical staff. Computer 12 or other digital processing device also includes or is operatively connected to one or more user input devices, such as an illustrative keyboard 16, or a mouse, trackball, trackpad, touch-sensitive screen. possibly integrated into the display device 14), or other pointer-based user input device, through which the average staff can input information such as operational commands to control the CDS system 10, data for use by the CDS system 10, and so on.
[0259] The CDS 10 system receives as input information regarding a medical material (for example, a hospital patient, or an outpatient treated by an oncologist, physician, or other medical team, or a person undergoing cancer screening or some other medical diagnosis known or suspected to have a certain type of cancer, such as colon cancer, breast cancer, or liver cancer, and so on). The CDS system 10 applies various data analysis algorithms to this input information to generate clinical decision support recommendations that are presented to the medical staff through the screen device 14 (or through a speech synthesizer or other device providing output noticeable to humans). In some embodiments, these algorithms may include applying a clinical guideline to the patient. A clinical guideline is a stored set of standard or “canonical” treatment recommendations, typically built on recommendations from a panel of medical experts and optionally formatted in the form of a clinical “flowchart” to facilitate navigation through the clinical guideline. In various embodiments, the CDS 10 data processing algorithms can additionally or alternatively include various diagnostic or clinical testing algorithms that are performed on input information to extract clinical decision recommendations, such as machine learning methods disclosed herein.
[0260] In the illustrative CDS systems disclosed in this document (for example, the CDS system 10), the CDS data analysis algorithms include one or more diagnostic or clinical test algorithms that are performed on genomic and/or proteomic information of entry acquired by one or more medical laboratories 18. These laboratories can be located in various ways, “on site”, ie, in the hospital or other location where the medical material is undergoing medical examination and/or treatment, or “outside of the site", for example, a specialized and centralized laboratory that receives (via mail or other delivery service) a tissue sample of the medical material that has been extracted from the medical material (for example, a sample obtained from a breast lesion, or from a colon from medical material known to, or suspected of having colon cancer, or a liver from medical material known to, or suspected of having liver cancer, or so on, through a biopsy procedure or other sample extraction procedure). The tissue from which a sample is taken may also be metastatic tissue, for example, tissue (suspected to be) malignant originating from the colon, breast, liver, or other organ that has spread outside the colon, breast, liver, or other organ. In some cases, the tissue sample can be circulating tumor cells, that is, tumor cells that have entered the bloodstream and can be extracted like the extracted tissue sample using appropriate isolation techniques. The extracted sample is processed by the laboratory to generate genomic or proteomic information. For example, the extracted sample can be processed using a microarray (also called in various ways in the art as a gene chip, DNA chip, biochip, and so on) or by quantitative polymerase chain reaction (qPCR) processing to measuring probative genomic or proteomic information, such as expression levels of genes of interest, for example, in the form of a level of messenger ribonucleic acid (mRNA) that is transcribed from the gene, or a level of protein that is translated from of the mRNA transcribed from the gene. As another example, the extracted sample can be processed by a gene sequencing laboratory to generate sequences for deoxyribonucleic acid (DNA), or to generate an RNA sequence, copy number variation, or so on. Other measurement approaches contemplated include immunohistochemistry (IHC), cytology, fluorescence in situ hybridization (FISH), proximity binding assay, and so on, performed on a pathology slide. Other information that can be generated by microarray processing, mass spectrometry, gene sequencing, or other laboratory techniques include methylation information. Various combinations of such genomic and/or proteomic measurements can also be performed.
[0261] In some achievements, medical laboratories 18 perform a series of standardized data acquisitions on the sample extracted from the tissue of medical material, in order to generate a large amount of genomic and/or proteomic data. For example, standard data acquisition techniques can generate a DNA sequence (optionally aligned) for one or more chromosomes or portions of chromosomes, or for the entire tissue genome. Applying a standard microarray can generate thousands or tens of thousands of data items, such as expression levels for a large number of genes, various methylation data, and so on. This plethora of genomic and/or proteomic data, or portions selected from them, are entered into the CDS 10 system to be processed in order to develop clinically useful information to formulate recommendations to support clinical decisions.
[0262] The disclosed CDS systems and disclosed methods refer to the processing of genomic and/or proteomic data to assess the activity of various cell signaling pathways. However, it should be understood that the disclosed CDS systems (eg, the CDS system 10) may optionally additionally include several additional capabilities, such as generating clinical decision support recommendations in accordance with stored clinical guidelines based on various patient data, such as vital signs monitoring data, patient history data, patient demographic data (for example, gender, age, and so on), patient medical imaging data, and so on. against. Alternatively, in some embodiments, the capabilities of the CDS 10 system may be limited to only performing genomic and/or proteomic data analysis to assess cell signaling pathways as disclosed herein.
[0263] With continued reference to exemplary Figure 20, the CDS 10 system infers the activity of a cell signaling pathway in tissue medical material based at least, but not limited to, on the expression levels of target genes of the pathway of cell signaling measured in the extracted sample, and determines whether the cell signaling pathway is operating abnormally in the tissue of the medical material based on this inferred activity. Examples disclosed herein refer to the Wnt, ER, AR and Hedgehog pathways as illustrative cell signaling pathways. These pathways are of interest in several areas of oncology because the loss of regulation in the pathways can be a cause of cancer proliferation. There are about 10 to 15 relevant signaling pathways, and each cancer is initially triggered by a dominant pathway that is unregulated. Without being limited to any specific theory of operation, these pathways regulate cell proliferation, and consequently, a loss of regulation of these pathways in cancer cells can lead to the pathway being "always on", thus accelerating the proliferation of cancer cells, which , in turn, manifests as cancer growth, invasion or metastasis (spread).
[0264] The measurement of mRNA expression levels of genes that code for regulatory proteins of the cell signaling pathway, such as an intermediate protein that is part of a cascade of proteins forming the cell signaling pathway, is an indirect measure of the expression level of the regulatory protein and may or may not strongly correlate with the level of expression of the regulatory protein (much less with the overall activity of the cell signaling pathway). The cell signaling pathway directly regulates transcription of target genes - thus, the expression levels of mRNA transcribed from target genes is a direct result of this regulatory activity. Thus, the CDS 10 system infers cell signaling pathway activity (eg, the Wnt, ER, AR and Hedgehog pathways) based at least on target gene expression levels (mRNA or protein level as a surrogate measure ) of the cell signaling pathway. This ensures that the CDS 10 system infers pathway activity based on direct information provided by measured expression levels of target genes.
[0265] However, although, as disclosed in this document, they are effective in assessing overall pathway activity, the measured expression levels of pathway target genes are not particularly informative as to why pathways are operating abnormally (if this really is the case). In other words, the measured expression levels of target genes in a pathway may indicate that the pathway is operating abnormally, but they do not indicate which portion of the pathway is malfunctioning (eg lacking sufficient regulation) to cause the general route operates abnormally.
[0266] Therefore, if the CDS 10 system detects abnormal activity of a specific pathway, the CDS 10 system then optionally makes use of other information provided by medical laboratories 18 for the extracted sample, such as aligned genetic sequences 22 and /or measured expression level(s) for one or more regulatory genes of pathway 24, or select the diagnostic test to be performed next to assess which portion of the pathway is malfunctioning. To maximize efficiency, in some embodiments, this optional assessment of why the pathway is malfunctioning is performed only if analysis of measured expression levels of the pathway's target genes indicates that the pathway is operating abnormally. In other embodiments, this assessment is integrated with the probabilistic analysis of the cell signaling pathway described in this document.
[0267] In embodiments in which the CDS 10 system assesses which portion of the pathway is malfunctioning, and is successful in doing so, the information allows the CDS 10 system to recommend the prescription of a drug targeted to the malfunction specific (recommendation 26 shown in Figure 20). If no specific roadway malfunctions are identified (because the optional additional assessment is not performed or because this assessment does not identify any specific portion of the roadway that is malfunctioning), then the CDS 10 system can provide a standard recommendation 28 recommending prescribing a general suppression drug for this specific pathway (assuming the abnormal pathway activity is very high activity).EXAMPLE 16: AN ANALYSIS KIT AND TOOLS TO MEASURE PATHWAY ACTIVITY
[0268] The set of target genes that are shown to best indicate specific pathway activity, based on investigation based on microarray/RNA sequencing using the Bayesian model, can be translated into a quantitative multiplex PCR assay to be performed on a sample of tissue or cell. To develop such an FDA-approved test for pathway activity, the development of a standardized test kit is required, which needs to be clinically validated in clinical trials to gain regulatory approval.
[0269] In general, it should be understood that, although samples belonging to the Wnt, ER, AR and/or Hedgehog pathway(s) are provided as illustrative examples, the approaches to cell signaling pathway analysis disclosed in this document they are readily applied to other cell signaling pathways in addition to these pathways, such as intracellular signaling pathways with receptors in the cell membrane (as above) and intracellular signaling pathways with receptors within the cell (as above). Furthermore: This patent application describes several preferred embodiments. Modifications and alterations may occur to others after reading and understanding the description detailed above. It is intended that the patent application be interpreted to include all modifications and alterations to the extent that they are within the scope of the appended claims or equivalents thereto.LITERATURE:
[0270] by Sousa E Melo F, C.S. (2011). Methylation of cancer-stem-cell-associated Wnt target genes predict poor prognosis in colorectal cancer patients. Cell Stem Cell., 476-485
[0271] Hatzis P, v. d. (2008). Genome-widepattern of TCF7L2/TCF4 chromatin occupancy in colorectal cancer cells. Mol Cell Biol., 2732-2744
[0272] Neapolitan, R. (2004). Learning Bayesiannetworks. Pearson Prentice Hall
[0273] Nusse, R. (2012, May 1). Wnt targetgenes. Retrieved from the Wnt homepage: http://www.stanford.edu/group/nusselab/cgi-bin/wnt/target_genes
[0274] Soderberg O, G.M. (2006). Direct observation of individual endogenous protein complexes in situ by proximity binding. Nat Methods., 995-1000
[0275] van de Wetering M, S.E.-P.-F. (2002).The beta-catenin/TCF-4 complex imposes on crypt parent

权利要求:
Claims (8)
[0001]
1. METHOD, characterized by comprising: the inference of the activity of one or more cell signaling pathway(s) in the tissue of a medical material, at least based on the expression level(s) (20) of one or plus target gene(s) of the cell signaling pathway(s) measured in a sample extracted from the tissue of medical material, where the inference comprises: the inference of the activity of the pathway(s) of cell signaling in the tissue of medical material by evaluating at least a portion of a probabilistic model (40-1, ..., 40-7), preferably a Bayesian network (40-1, ..., 40-7) , representing the cell signaling pathway(s) for a set of inputs including at least the expression level(s) (20) of one or more target genes of the signaling pathway(s) cell measurements in the extracted tissue medical material sample; the estimation of a level (46) in the medical material tissue of at least one transcription factor (TF) element, the at least one TF element controlling the transcription of one or more target gene(s) from the cell signaling pathway(s), the estimate being based, at least in part, on conditional probabilities regarding at least one TF element and the expression level(s) (20) of the one or more target gene(s) of the cell signaling pathway(s) measured in the sample extracted from the tissue of the medical material; the inference of the activity of the ) cell signaling pathway(s) based on the estimated level of transcription factor in the tissue sample; and determining the cell signaling pathway(s) to be/are or not operating abnormally in the tissue of the medical device based on the inferred activity of the cell signaling pathway(s) in the tissue of the medical device; in which the inference is performed by a digital processing device (12) using the probabilistic model (40-1, ..., 40-7) of the cell signaling pathway(s), in which the pathway(s) s) cell signaling is/are comprises a Wnt pathway, an ER pathway, an AR pathway and/or a Hedgehog pathway, where the inference comprises: the inference of the activity of the Wnt pathway in the tissue of medical material based on at least (20) expression levels of one or more, preferably at least three, Wnt pathway target gene(s) measured in the tissue-extracted sample of medical material selected from the group consisting of: KIAA1199, AXIN2 , RNF43, TBX3, TDGF1, SOX9, ASCL2, IL8, SP5, ZNRF3, KLF6, CCND1, DEFA6 and FZD7, and/or the inference of ER pathway activity in the tissue of the mother material based on at least (20) expression levels of one or more, preferably at least three, target gene(s) of the ER pathway measured in the extracted tissue sample of medical material selected from the group consisting of: CDH26 , SGK3, PGR, GREB1, CA12, XBP1, CELSR2, WISP2, DSCAM, ERBB2, CTSD, TFF1 and NRIP1, and/or inference of Hedgehog pathway activity in tissue medical material based at least on expression levels (20) of one or more, preferably at least three, Hedgehog pathway target gene(s) measured in the tissue-extracted sample of medical material selected from the group consisting of: GLI1, PTCH1, PTCH2, IGFBP6, SPP1, CCND2, FST, FOXL1 , CFLAR, TSC22D1, RAB34, S100A9, S100A7, MYCN, FOXM1, GLI3, TCEA2, FYNe CTSL1, and/or the inference of AR pathway activity in tissue from medical material based on at least (20) expression levels of one or more, preferably at least three, AR pathway target gene(s) measured in the tissue sample extracted from the medical material if taught from the group consisting of: KLK2, PMEPA1, TMPRSS2, NKX3_1, ABCC4, KLK3, FKBP5, ELL2, UGT2B15, DHCR24, PPAP2A, NDRG1, LRIG1, CREB3L4, LCP1, GUCY1A3, AR and EAF2.
[0002]
2. METHOD, according to claim 1, characterized in that the inference comprises: the estimation of a level (46) in the tissue of the medical material of at least one element of transcription factor (TF), represented by a TF node of the model probabilistic, the TF element controlling the transcription of one or more target gene(s) of the cell signaling pathway(s), the estimate being based, at least in part, on conditional probabilities of the probabilistic model (40- 1, ..., 40-7) relating the TF node and the nodes in the probabilistic model representing the one or more target gene(s) of the cell signaling pathway(s) measured in the tissue sample extracted from the material medical, and wherein the inference is preferably performed using a Bayesian network (40-1, ., 40-7) comprising nodes representing information about the signaling pathway(s) and conditional probability relations between connected nodes of the network Bayesian.
[0003]
3. METHOD according to claim 1 or 2, characterized in that the inference is additionally based on expression levels (20) of at least one target gene of the Wnt pathway measured in the sample extracted from the tissue of the medical material selected from the group comprising: NKD1, OAT, FAT1, LEF1, GLUL, REG1B, TCF7L2, COL18A1, BMP7, SLC1A2, ADRA2C, PPARG, DKK1, HNF1A and LECT2.
[0004]
4. METHOD according to claim 1 or 2, characterized in that the inference is additionally based on expression levels (20) of at least one target gene of the ER pathway measured in the sample extracted from the tissue of the medical material selected from the group comprising: AP1B1, ATP5J, COL18A1, COX7A2L, EBAG9, ESR1, HSPB1, IGFBP4, KRT19, MYC, NDUFV3, PISD, PRDM15, PTMA, RARA, SOD1 and TRIM25.
[0005]
5. METHOD according to claim 1 or 2, characterized in that the inference is additionally based on expression levels (20) of at least one target gene of the Hedgehog pathway measured in the sample extracted from the tissue of the medical material selected from the group comprising: BCL2, FOXA2, FOXF1, H19, HHIP, IL1R2, JAG2, JUP, MIF, MYLK, NKX2-2, NKX2-8, PITRM1 and TOM1.
[0006]
6. METHOD according to claim 1 or 2, characterized in that the inference is additionally based on expression levels (20) of at least one target gene of the RA pathway measured in the sample extracted from the tissue of the medical material selected from the group comprising: APP, NTS, PLAU, CDKN1A, DRG1, FGF8, IGF1, PRKACB, PTPN1, SGK1 and TACC2.
[0007]
7. APPARATUS, characterized in that it comprises a digital processor (12) configured to carry out a method as defined in any one of claims 1 to 6.
[0008]
8. NON-TRANSITARY STORAGE MEANS, characterized by storing instructions that are executable by a digital processing device (12) to perform a method as defined in any one of claims 1 to 6.

类似技术:

---

公开号 | 公开日 | 专利标题

---

JP6807430B2|2021-01-06|Evaluation of cellular signaling pathway activity using a probabilistic model of target gene expression

---

JP2019076096A|2019-05-23|Assessment of cellular signaling pathway activity using linear combinations of target gene expressions

---

AU2014259459B2|2020-03-26|Medical prognosis and prediction of treatment response using multiple cellular signalling pathway activities

同族专利:

---

公开号 | 公开日

---

EP2734643A2|2014-05-28|

---

CN109852671A|2019-06-07|

---

EP3418396B1|2021-03-24|

---

JP2017205129A|2017-11-24|

---

WO2013011479A3|2013-03-14|

---

JP6204354B2|2017-09-27|

---

CN103649337B|2018-11-13|

---

DK3418396T3|2021-05-25|

---

JP2014520566A|2014-08-25|

---

RU2014106026A|2015-08-27|

---

KR102064004B1|2020-01-08|

---

EP3831957A1|2021-06-09|

---

DK2734643T3|2017-03-20|

---

EP3173489A1|2017-05-31|

---

CN103649337A|2014-03-19|

---

EP3418396A1|2018-12-26|

---

WO2013011479A2|2013-01-24|

---

JP6783729B2|2020-11-11|

---

EP2549399A1|2013-01-23|

---

BR112014000965A2|2017-06-13|

---

MX352827B|2017-12-11|

---

IN2014CN00675A|2015-04-03|

---

MX2014000557A|2014-05-01|

---

EP3173489B1|2018-09-19|

---

RU2719194C2|2020-04-17|

---

US20140156200A1|2014-06-05|

---

KR20140046464A|2014-04-18|

---

JP2019129850A|2019-08-08|

---

JP6807430B2|2021-01-06|

---

ES2874452T3|2021-11-05|

---

EP2734643B1|2016-12-07|

---

ES2617195T3|2017-06-15|

---

ES2700617T3|2019-02-18|

---

DK3173489T3|2018-12-10|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US7371736B2|2001-11-07|2008-05-13|The Board Of Trustees Of The University Of Arkansas|Gene expression profiling based identification of DKK1 as a potential therapeutic targets for controlling bone loss|

---

AU2003260973A1|2002-09-05|2004-03-29|Toolgen, Inc.|Bioinformatics analysis of cellular effects of artificial transcription factors|

---

EP1842147A2|2005-01-24|2007-10-10|The Board of Trustees of The Leland Stanford Junior University|Method for modeling cell signaling systems by means of bayesian networks|

---

US20090186024A1|2005-05-13|2009-07-23|Nevins Joseph R|Gene Expression Signatures for Oncogenic Pathway Deregulation|

---

WO2007136787A2|2006-05-17|2007-11-29|The Trustees Of Boston University|Method to determine transcriptional regulation pathways in organisms|

---

JP2010178650A|2009-02-04|2010-08-19|Univ Of Tokyo|Test method for predicting recurrence of solid cancer and recurrence prophylactic|

---

JP6355555B2|2011-04-01|2018-07-11|キアゲン|Gene expression signature for WNT / B-catenin signaling pathway and use thereof|

---

CN103874768A|2011-05-05|2014-06-18|临床基因组学股份有限公司|A method of diagnosing neoplasms|RU2636000C2|2012-02-28|2017-11-17|Новартис Аг|Selection of cancer patients for introduction of wnt signal inhibitors based on rnf43 mutation status|

---

CA2881394A1|2012-08-13|2014-02-20|Cedars-Sinai Medical Center|Exosomes and micro-ribonucleic acids for tissue regeneration|

---

BR112015015131A2|2012-12-26|2017-07-11|Koninklijke Philips Nv|method, device, non-transient storage media, computer program, and product|

---

TWI582239B|2013-03-11|2017-05-11|諾華公司|Markers associated with wnt inhibitors|

---

EP2989211B1|2013-04-26|2018-06-13|Koninklijke Philips N.V.|Medical prognosis and prediction of treatment response using multiple cellular signalling pathway activities|

---

CA2919768C|2013-06-28|2019-12-03|Nantomics, Llc|Pathway analysis for identification of diagnostic tests|

---

US20160237163A1|2013-10-02|2016-08-18|Suri Technologies Ltd.|Patient-specific immunotherapy for treating heterogeneous tumors|

---

CN105492630A|2014-01-03|2016-04-13|皇家飞利浦有限公司|Assessment of the pi3k cellular signaling pathway activity using mathematical modelling of target gene expression.|

---

US9838847B2|2014-09-11|2017-12-05|Google LLP|Data driven evaluation and rejection of trained Gaussian process-based wireless mean and standard deviation models|

---

US9880257B2|2014-09-11|2018-01-30|Google Llc|Gaussian process-based approach for identifying correlation between wireless signals|

---

US9810762B2|2014-09-11|2017-11-07|Google Inc.|Calculating mean wireless signal strengths using a gaussian process approach incorporating predictive standard deviations|

---

WO2016042164A1|2014-09-19|2016-03-24|The Provost, Fellows, Foundation Scholars, & The Other Members Of Board, Of The College Of The Holy & Unidv. Trinity Of Queen Elizabeth, Near Dublin|A method of predicting risk of recurrence of cancer|

---

BR112017007965A2|2014-10-24|2018-01-23|Koninklijke Philips Nv|computer-implemented method for inferring the activity of a tgf-b cell signaling pathway in an individual; apparatus for inferring the activity of a tgf-b cellular signaling pathway in an individual; non-transient storage medium; computer program; kits for measuring expression levels of three or more tgf-b cell signaling pathway target genes in a sample from an individual; infer the activity of a tgf-b cell signaling pathway in an individual; infer the activity of a tgf-b cell signaling pathway in an individual; and use of the kit|

---

WO2016062893A1|2014-10-24|2016-04-28|Koninklijke Philips N.V.|Medical prognosis and prediction of treatment response using multiple cellular signaling pathway activities|

---

WO2016062892A1|2014-10-24|2016-04-28|Koninklijke Philips N.V.|Medical prognosis and prediction of treatment response using multiple cellular signaling pathway activities|

---

CN105760707B|2014-12-18|2018-07-31|中国科学院大连化学物理研究所|Cytogene translation process modeling method|

---

CA2995520A1|2015-08-14|2017-02-23|Hendrik Jan VAN OOIJEN|Assessment of nfkb cellular signaling pathway activity using mathematical modelling of target gene expression|

---

WO2017033154A1|2015-08-27|2017-03-02|Koninklijke Philips N.V.|An integrated method and system for identifying functional patient-specific somatic aberations using multi-omic cancer profiles|

---

CN106897578B|2015-12-15|2020-02-14|中国科学院大连化学物理研究所|Modeling method for cell gene translation process|

---

EP3402543B1|2016-01-11|2021-09-08|Cedars-Sinai Medical Center|Cardiosphere-derived cells and exosomes secreted by such cells in the treatment of heart failure with preserved ejection fraction|

---

US20170329914A1|2016-05-11|2017-11-16|International Business Machines Corporation|Predicting Personalized Cancer Metastasis Routes, Biological Mediators of Metastasis and Metastasis Blocking Therapies|

---

US10621236B2|2016-09-16|2020-04-14|At&T Intellectual Property I, L.P.|Concept based querying of graph databases|

---

EP3538673A4|2016-11-11|2019-12-04|University of Pittsburgh- Of the Commonwealth System of Higher Education|Identification of instance-specific somatic genome alterations with functional impact|

---

CA3044709A1|2016-11-25|2018-05-31|Koninklijke Philips N.V.|Method to distinguish tumor suppressive foxo activity from oxidative stress|

---

CN108300783A|2017-01-11|2018-07-20|上海易毕恩基因科技有限公司|The method of gene marker for screening intestinal cancer and/or gastric cancer, the gene marker and application thereof screened with this method|

---

CN106755464A|2017-01-11|2017-05-31|上海易毕恩基因科技有限公司|For the method for screening the gene marker of intestinal cancer and/or stomach cancer, the gene marker and application thereof that is screened with the method|

---

CN107184991B|2017-05-03|2020-05-26|武汉爱博泰克生物科技有限公司|Application of PRDM15 gene sequence in preparation of NF-kB inhibitor|

---

CN107164508A|2017-06-16|2017-09-15|上海易毕恩基因科技有限公司|Gene marker for detecting liver cancer and application thereof|

---

EP3431582A1|2017-07-18|2019-01-23|Koninklijke Philips N.V.|Cell culturing materials|

---

CN107201411A|2017-07-27|2017-09-26|上海市长宁区妇幼保健院|MYLK genes as diagnosis of endometrial carcinoma mark|

---

CN107385051A|2017-08-04|2017-11-24|上海易毕恩生物技术有限公司|For detecting liver tumour good pernicious gene marker, kit and detection method|

---

EP3462348A1|2017-09-28|2019-04-03|Koninklijke Philips N.V.|Bayesian inference|

---

EP3461916A1|2017-10-02|2019-04-03|Koninklijke Philips N.V.|Assessment of jak-stat3 cellular signaling pathway activity using mathematical modelling of target gene expression|

---

EP3461915A1|2017-10-02|2019-04-03|Koninklijke Philips N.V.|Assessment of jak-stat1/2 cellular signaling pathway activity using mathematical modelling of target gene expression|

---

US20200240979A1|2017-10-02|2020-07-30|Koninklijke Philips N.V.|Determining functional status of immune cells types and immune response|

---

EP3462349A1|2017-10-02|2019-04-03|Koninklijke Philips N.V.|Assessment of notch cellular signaling pathway activity using mathematical modelling of target gene expression|

---

EP3502279A1|2017-12-20|2019-06-26|Koninklijke Philips N.V.|Assessment of mapk-ap 1 cellular signaling pathway activity using mathematical modelling of target gene expression|

---

CN113626397A|2018-06-16|2021-11-09|王梅|Method and system for dynamically processing data file|

---

WO2020225155A1|2019-05-03|2020-11-12|Koninklijke Philips N.V.|Methods of prognosis in high-grade serous ovarian cancer|

---

AU2020267561A1|2019-05-08|2021-12-23|Capricor, Inc.|Therapeutically active cells and exosomes|

---

EP3739588A1|2019-05-13|2020-11-18|Koninklijke Philips N.V.|Assessment of multiple signaling pathway activity score in airway epithelial cells to predict airway epithelial abnormality and airway cancer risk|

---

EP3812474A1|2019-10-22|2021-04-28|Koninklijke Philips N.V.|Methods of prognosis in high-grade serous ovarian cancer|

---

EP3822368A1|2019-11-14|2021-05-19|Koninklijke Philips N.V.|Assessment of pr cellular signaling pathway activity using mathematical modelling of target gene expression|

---

WO2021119641A1|2019-12-12|2021-06-17|Tempus Labs, Inc.|Real-world evidence of diagnostic testing and treatment patterns in u.s. breast cancer patients|

---

EP3835433A1|2019-12-12|2021-06-16|Koninklijke Philips N.V.|Notch signaling pathway activity as prognostic marker in bladder cancer|

---

CN111139300B|2020-02-19|2021-08-17|伯克利南京医学研究有限责任公司|Application of group of colon cancer prognosis related genes|

---

EP3882363A1|2020-03-17|2021-09-22|Koninklijke Philips N.V.|Prognostic pathways for high risk sepsis patients|

---

WO2021209567A1|2020-04-16|2021-10-21|Koninklijke Philips N.V.|Prognostic pathways for viral infections|

---

EP3940704A1|2020-07-14|2022-01-19|Koninklijke Philips N.V.|Method for determining the differentiation state of a stem cell|

---

EP3960875A1|2020-08-28|2022-03-02|Koninklijke Philips N.V.|Pcr method and kit for determining pathway activity|

---

EP3965119A1|2020-09-04|2022-03-09|Koninklijke Philips N.V.|Methods for estimating heterogeneity of a tumour based on values for two or more genome mutation and/or gene expression related parameter, as well as corresponding devices|

---

CN113278700A|2021-06-04|2021-08-20|浙江省肿瘤医院|Primer group and kit for breast cancer typing and prognosis prediction|

法律状态:
2020-04-07| B06U| Preliminary requirement: requests with searches performed by other patent offices: procedure suspended [chapter 6.21 patent gazette]|

---

2021-06-01| B350| Update of information on the portal [chapter 15.35 patent gazette]|

---

2021-06-15| B09A| Decision: intention to grant [chapter 9.1 patent gazette]|

---

2021-08-24| B16A| Patent or certificate of addition of invention granted [chapter 16.1 patent gazette]|Free format text: PRAZO DE VALIDADE: 20 (VINTE) ANOS CONTADOS A PARTIR DE 19/07/2012, OBSERVADAS AS CONDICOES LEGAIS. |

优先权:

---

申请号 | 申请日 | 专利标题

---

US201161509137P| true| 2011-07-19|2011-07-19|

---

US61/509,137|2011-07-19|

---

EP11178148.0|2011-08-19|

---

EP11178148A|EP2549399A1|2011-07-19|2011-08-19|Assessment of Wnt pathway activity using probabilistic modeling of target gene expression|

---

PCT/IB2012/053686|WO2013011479A2|2011-07-19|2012-07-19|Assessment of cellular signaling pathway activity using probabilistic modeling of target gene expression|

---