Inference of patient-specific pathway activities from multi-dimensional cancer genomics data using PARADIGM

摘要: Motivation: High-throughput data is providing a comprehensive view of the molecular changes in cancer tissues. New technologies allow for simultaneous genome-wide assay state genome copy number variation, gene expression, DNA methylation and epigenetics tumor samples cell lines. Analyses current sets find that genetic alterations between patients can differ but often involve common pathways. It therefore critical to identify relevant pathways involved progression detect how they are altered different patients. Results: We present novel method inferring patient-specific activities incorporating curated pathway interactions among genes. A modeled by factor graph as set interconnected variables encoding expression known activity its products, allowing incorporation many types omic evidence. The predicts degree which pathway's (e.g. internal states, or high-level ‘outputs’) patient using probabilistic inference. Compared with competing inference approach called SPIA, our identifies cancer-related fewer false-positives both glioblastoma multiform (GBM) breast dataset. PARADIGM identified consistent pathway-level subsets GBM overlooked when genes considered isolation. Further, grouping based on their significant perturbations divides them into clinically-relevant subgroups having significantly survival outcomes. These findings suggest therapeutics might be chosen target at points commonly perturbed pathway(s) group patients. Availability:Source code available http://sbenz.github.com/Paradigm Contact: jstuart@soe.ucsc.edu Supplementary information:Supplementary Bioinformatics online.