Comprehensive analysis of mutational processes across 20 000 adult and pediatric tumors.

Despite being a consolidated practice in modern cancer genomics, mutational signature analysis poses various challenges. First, determining the number of signatures is a complex task and depends on heuristics. Second, several signatures lack a clear etiology, raising concerns about whether they result from computational artifacts or actual mutagenic processes. Last, approaches for signature assignment are highly influenced by the catalogue of signatures used for the analysis. To overcome these limitations, we introduce RESOLVE (Robust EStimation Of mutationaL signatures Via rEgularization), a framework designed for the efficient extraction, assignment, and confidence estimation of mutational signatures. RESOLVE enables the stratification of cancer genomes according to the active mutational processes, providing insights into those that are consistently shared among various cancer types. We applied RESOLVE to 20 000 samples from adult and pediatric cancers, providing a comprehensive characterization of subtypes associated with specific mutational processes and alterations in driver genes. Our analysis demonstrates that RESOLVE accurately fits observed mutations with a smaller set of signatures compared to existing catalogues, suggesting the existence of a limited number of dominant mutational processes in cancer genomes. Clustering analysis revealed distinct patient groups characterized by specific signatures, with significant associations between certain signatures and patient prognosis. Additionally, we identified strong associations between signatures and driver gene mutations, offering insights into cancer subtype mechanisms and evolution. Our findings highlight the efficiency of RESOLVE and its potential impact on personalized cancer treatment.