Key points are not available for this paper at this time.
Abstract The development of mutational signatures defined individually in the contexts of single/double base substitutions, insertions and deletions, structural variants and copy number variation have led to improved understanding of cancer mutagenesis and potential for improved patient stratification. We hypothesized that combining these contexts into a multi-modal mutational signature framework may improve the accuracy of assigning mutational processes to samples, and provide a holistic view of the full repertoire of mutations conferred from such processes. In order to develop multi-modal signatures, bespoke feature engineering was required for each mutational modality to enhance signal, in addition to individual modality matrix normalization so that the data could be harmonized and analyzed as a whole. These methods, and subsequent signature extraction were developed on 2, 438 whole genome sequenced cancers from 21 distinct cancer types from the PCAWG consortium for which all modalities of data were available, incorporating a 348 channel signature definition. Deployment of these methods identified multi-modal signatures that encompass multiple mutation types known to associate with specific processes, for example SBS, indel and rearrangement channels defining a signature of homologous recombination deficiency. In contrast, some signatures remain defined by a single mutational modality, highlighting the distinction between broadly mutational processes, and more focused mutation processes. Multi-modal signatures show promise in increasing accuracy for difficult to assign signatures, such as those for homologous recombination deficiency, and therefore in improving patient stratification for life-saving treatment. Citation Format: Christopher D. Steele, Ludmil Alexandrov, Azhar Khandekar. Multi-modal mutational signatures elucidate the comprehensive consequences of mutational processes abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts) ; 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84 (6Suppl): Abstract nr 3563.
Steele et al. (Fri,) studied this question.