BaySyn: Bayesian Evidence Synthesis for Multi-system Multiomic Integration

The discovery of cancer drivers and drug targets are often limited to the biological systems - from cancer model systems to patients. While multiomic patient databases have sparse drug response data, cancer model systems databases, despite covering a broad range of pharmacogenomic platforms, provide lower lineage-specific sample sizes, resulting in reduced statistical power to detect both functional driver genes and their associations with drug sensitivity profiles. Hence, integrating evidence across model systems, taking into account the pros and cons of each system, in addition to multiomic integration, can more efficiently deconvolve cellular mechanisms of cancer as well as learn therapeutic associations. To this end, we propose BaySyn - a hierarchical Bayesian evidence synthesis framework for multi-system multiomic integration. BaySyn detects functionally relevant driver genes based on their associations with upstream regulators using additive Gaussian process models and uses this evidence to calibrate Bayesian variable selection models in the (drug) outcome layer. We apply BaySyn to multiomic cancer cell line and patient datasets from the Cancer Cell Line Encyclopedia and The Cancer Genome Atlas, respectively, across pan-gynecological cancers. Our mechanistic models implicate several relevant functional genes across cancers such as PTPN6 and ERBB2 in the KEGG adherens junction gene set. Furthermore, our outcome model is able to make higher number of discoveries in drug response models than its uncalibrated counterparts under the same thresholds of Type I error control, including detection of known lineage-specific biomarker associations such as BCL11A in breast and FGFRL1 in ovarian cancers. All our results and implementation codes are freely available via an interactive R Shiny dashboard.