Foundation-model-guided radiogenomic discovery linking cancer genomes to cancer scans

2026-07-22

AI Summary

Traditional approaches to driver gene discovery rely on high-frequency mutations and struggle to identify rare yet functionally significant drivers. This study pioneers the integration of Evo²—a task-agnostic, pre-trained genomic language model—with clinical radiomics to predict the pathogenicity of somatic mutations across the whole genome without prior assumptions, while linking these predictions to imaging phenotypes. By incorporating tumor segmentation and rigorous multiple hypothesis testing correction (FDR control), the method not only recapitulates known driver genes in clear cell renal cell carcinoma but also identifies 46 novel significantly associated genes absent from current cancer gene panels. These newly implicated genes enrich pathways related to ciliopathies and cytoskeletal regulation, substantially expanding the landscape of rare driver gene discovery.

Abstract

The function of many genes is still unknown, and conventional driver-discovery methods, which rely on how frequently a gene is mutated, cannot assess genes that are only rarely affected. Here we pair Evo~2-based genome analysis with routine clinical imaging to identify gene--phenotype associations at genome-wide scale. For every somatic mutation across three TCGA cohorts (cRCC=clear cell renal cell carcinoma, HCC=hepatocellular carcinoma, and BC=breast cancer; $n = 340$ total), Evo~2 predicts a severity score, with no task-specific training. Per-gene severity summaries are then correlated with radiomic features extracted from paired tumor segmentations, controlling for total mutation burden. In TCGA-cRCC ($n = 162$), this sweep recovers established renal-cancer drivers and identifies 46 additional genes reaching false discovery rate (FDR) significance absent from curated cancer-gene panels, several of which are Mendelian ciliopathy and cytoskeletal-disease genes. These results demonstrate that pairing a genomic language model with widely available clinical imaging can serve as a hypothesis-free discovery tool for gene--imaging associations invisible to conventional approaches.

Problem

• radiogenomics
• rare mutations
• gene-phenotype association
• cancer genomics
• clinical imaging

Innovation

• foundation model
• radiogenomics
• Evo2
• gene-phenotype association
• hypothesis-free discovery