TASTE identifies shared proteomic effects on multiple related cancers

IntroductionGenome-wide association studies (GWAS) have identified hundreds of variants linked to cancers, but their downstream regulatory consequences remain poorly understood. Increasing evidence suggests that related cancers share alterations of common regulatory programs. Trans-associations of cancer risk variants mediated via molecular phenotypes, such as gene expression and protein levels, can help uncover these downstream mechanisms. Further investigation of such convergence can reveal shared etiological processes and therapeutic targets across cancer types. MethodWe introduce TASTE (Trans Association using Shared factorization and TEsting), a summary statistic-based framework to identify protein sets that are trans-regulated by genetic variants associated with sets of biologically related cancers. TASTE consists of three steps: (1) TASTE-D, a low-rank matrix factorization to estimate shared and group-specific trans-association patterns across cancers; (2) TASTE-S, a sparse singular value decomposition to identify proteins driving shared effects; (3) TASTE-T, a competitive testing strategy for evaluating significance of trans-associations captured by the identified protein-set. ResultsSimulation studies show that TASTE can identify protein-set with shared effects across groups of related cancers with high accuracy. Applying TASTE to UK Biobank plasma proteomic data and GWAS of genitourinary cancers (renal cell carcinoma [RCC] and bladder cancer [BLCA]), we identified 88 proteins with shared trans-associations. Notably, 43% of these proteins lacked significant trans-pQTL but were discovered through aggregation of weak signals only. This includes KDR (p=1.4x10-7), a VEGF receptor and drug target for RCC and BLCA, nominally associated with 10 RCC and 8 BLCA loci. Additional in silico analyses identified enrichment in cancer-relevant pathways such as PI3K/AKT, MAPK, VEGF, and Ras signaling, as well as significant protein-protein interaction and differential expression in tumors. Identified proteins were overrepresented in known cancer drivers and genes essential to cell proliferation in RCC and BLCA cell lines. ConclusionTASTE is a robust and scalable approach to uncover convergent trans-regulatory mechanisms across related cancers or diseases enabling discovery of downstream molecular targets and highlighting candidates for mechanistic follow-up and therapeutic development.