Topographic modeling of mutational signatures across the cancer genome
MuTopia learns topographic models of somatic mutation — it simultaneously decomposes a cancer cohort's mutations into signatures and explains how local genomic context (chromatin state, replication timing, transcription level, DNA sequence composition) shapes each signature's activity across 10 kb windows genome-wide. Provides G-Tensor data structures, SHAP-based feature interpretability, and per-mutation VCF annotation.
