Molecular pathways: extracting medical knowledge from high-throughput genomic data

Abstract

High-throughput genomic data that measures RNA expression, DNA copy number, mutation status, and protein levels provide us with insights into the molecular pathway structure of cancer. Genomic lesions (amplifications, deletions, mutations) and epigenetic modifications disrupt biochemical cellular pathways. Although the number of possible lesions is vast, different genomic alterations may result in concordant expression and pathway activities, producing common tumor subtypes that share similar phenotypic outcomes. How can these data be translated into medical knowledge that provides prognostic and predictive information? First-generation mRNA expression signatures such as Genomic Health's Oncotype DX already provide prognostic information, but do not provide therapeutic guidance beyond the current standard of care, which is often inadequate in high-risk patients. Rather than building molecular signatures based on gene expression levels, evidence is growing that signatures based on higher-level quantities such as from genetic pathways may provide important prognostic and diagnostic cues. We provide examples of how activities for molecular entities can be predicted from pathway analysis and how the composite of all such activities, referred to here as the "activitome," helps connect genomic events to clinical factors to predict the drivers of poor outcome.

Figure 1. PARADIGM model for integrative data analysis

A. Factor graph model oriented around a single gene Hidden states in a tumor sample (open ellipses) for genomic copies (G), epigenetic promoter state (E), mRNA transcripts (T), peptide (P), and active protein (A). Regulation gene expression (open ellipses) include transcriptional (R_T), translational (R_P), and post-translational (R_A) control. Sample data (filled circles, constrain gene states through factors (boxes). B. Toy example of a MYC/MAX-associated pathway. Two transcription factors (MYC and MAX) form a complex (MYC/MAX) that is inhibited by PAK2, a protein kinase. MYC/MAX activates two target genes (CCNB1, ENO1) and inactivates a third (WNT5A). C. Single patient data converted to inferred activities for toy pathway. Measurements and inferred levels are either higher (red), lower (blue), or comparable (purple) to levels in matched normal. Belief propagation infers the kinase is inactive based on inferred higher activity of MYC/MAX.

Figure 2. HotLink Result for TCGA Breast Cancer

Rings of data depict different measurements about genes or proteins as higher activity (red) or lower activity (blue) compared to normal controls. Pathway illustrates part of the HotLink solution inferred from the TCGA basal and luminal breast tumors with various data available through TCGA including (inner to outer): pathway levels inferred by PARADIGM, Copy number alterations, RNA-Seq RSEM levels, and RPPA data. The outermost ring depicts the patient subtypes. Segments display aggregated levels for samples within each grouping defined by the breast cancer subtype (indicated in the outermost ring).

Figure 3

DiPSC(Dipstick) depicts correlations comparing mutations, and biomarkers along the Luminal-A/Luminal B dichotomy.