A pan-system approach for mutational signature identification and etiology assignment in cancer genomics

Purpose : Somatic mutations accumulate in the genomes of both normal and cancer cells. Mutagens imprint characteristic genomic patterns of somatic mutations, termed, mutational signatures. Accurate identification of mutational signatures is essential for linking genotoxic exposures and endogenous processes to cancer etiology. However, technical artefacts, sequencing noise, and platform heterogeneity often compromise the fidelity of signature detection, limiting mechanistic insights and biomarker discovery. Methods : We developed SigRescueR , a pan-system computational tool that performs noise correction and robust mutational signature identification across diverse sequencing platforms. The framework integrates adaptive artefact filtering, robust signature reconstruction, and cosine similarity-based signal rescue across diverse mutational classes, including single base substitutions (SBS), insertions and deletions (indels), doublet base substitutions (DBS), and strand-bias mutation categories. Results : We applied SigRescueR to the largest collection of experimentally derived mutational profiles assembled to date, encompassing over 4,200 genome-wide sequencing profiles across 42 model systems exposed to 120+ mutagenic agents under standardized experimental conditions. Comparative analysis across human, mouse, rat, chicken, and C. elegans models revealed 49 robust experimental mutational signatures. Several experimentally derived signatures reflected conserved DNA damage patterns across species and models exposed to the same agent or related chemicals, while others were species- or model-specific, suggesting system-dependent variability in mutational response. Moreover, several experimental mutational signatures corresponded to established COSMIC patterns, while others represented previously uncharacterized DNA damage processes in human cancers. Importantly, this compendium enabled causal assignment of previously unexplained COSMIC signatures, namely SBS94 and SBS95, to defined environmental exposures. Conclusion: Together, SigRescueR and this compendium provide a unified, high-resolution view of mutagenesis, connecting molecular toxicology, cancer genomics, and biomarker discovery. This integrative framework refines how we trace, interpret, and incite measures to prevent the genomic imprints of carcinogenesis.