A Compendium of 49 Experimental SBS Signatures for Decoding Human Cancer Mutational Processes

Abstract

Human cancer genomes harbor distinct mutational patterns that reflect past processes of DNA damage and repair. However, the precise attribution of these signatures to specific chemical carcinogens lacks a standardized experimental reference framework. To address this gap, we curated 4,282 genome-wide sequencing datasets from 42 model systems across five species exposed to 146 cancer-risk agents. This platform yielded 49 robust experimental single-base substitution signatures (eSS), with 28 matching 19 established COSMIC signatures and 21 defining novel mutational processes. We reconstructed 24 COSMIC signatures, assigning candidate etiologies to five signatures of unknown origin and revising two contested assignments. Pan-cancer decomposition detected four eSS-like mutational processes enriched in smokers across 4,951 tumors. Lastly, independent single-molecule sequencing of primary human organoids reproduced these profiles with high fidelity, confirming true platform-independent biological reproducibility across complex human models. This eSS repertoire provides a reference that links human mutational processes to mechanistic classes of DNA damage.