A comprehensive comparison of tools for fitting mutational signatures

Abstract

Mutational signatures connect characteristic mutational patterns in the genome with biological or chemical processes that take place in cancers. Analysis of mutational signatures can help elucidate tumor evolution, prognosis, and therapeutic strategies. Although tools for extracting mutational signatures de novo have been extensively benchmarked, a similar effort is lacking for tools that fit known mutational signatures to a given catalog of mutations. We fill this gap by comprehensively evaluating twelve signature fitting tools on synthetic mutational catalogs with empirically driven signature weights corresponding to eight cancer types. On average, SigProfilerSingleSample and SigProfilerAssignment/MuSiCal perform best for small and large numbers of mutations per sample, respectively. We further show that ad hoc constraining the list of reference signatures is likely to produce inferior results. Evaluation of real mutational catalogs suggests that the activity of signatures that are absent in the reference catalog poses considerable problems to all evaluated tools.

Various biological and chemical processes leave characteristic patterns, mutational signatures, in the genome. Here the authors assess tools for fitting known mutational signatures to sequenced samples (to determine the signature contributions to each individual sample), finding that they are all prone to underfitting due to the activity of unknown signatures.