#36 · Genomics

Causal, kernel, and multi-trait methods extend genetic association testing

The dominant thread is extending genetic association analysis beyond single-trait, single-context designs. METEOR jointly models multiple outcomes from summary statistics while accounting for sample overlap, and MULTI pools genetic instruments across tissues to identify causal disease risk genes. HiC-KAT incorporates 3D genome structure by weighting variant-set similarity with Hi-C contact confidence for gene-agnostic rare-variant testing in WGS data. InvSE-MTAGR and InvPSE-MTAGR apply Shannon-entropy-based multi-trait gene-region association, with simulations reporting Type I error control and power. Sparse CCA mediation handles high-dimensional exposures and mediators, illustrated with a metabolite pathway application to MELD score. WDCN, a deep network for breast cancer risk, is predictive rather than inferential, modeling gene-gene and gene-environment interactions with AUROC up to 0.887 relative to PRS.