Integrated structural variation and point mutation signatures in cancer genomes using correlated topic models

Abstract

Mutation signatures in cancer genomes reflect endogenous and exogenous mutational processes, offering insights into tumour etiology, features for prognostic and biologic stratification and vulnerabilities to be exploited therapeutically. We present a novel machine learning formalism for improved signature inference, based on multi-modal correlated topic models (MMCTM) which can at once infer signatures from both single nucleotide and structural variation counts derived from cancer genome sequencing data. We exemplify the utility of our approach on two hormone driven, DNA repair deficient cancers: breast and ovary (n = 755 samples total). We show how introducing correlated structure both within and between modes of mutation can increase accuracy of signature discovery, particularly in the context of sparse data. Our study emphasizes the importance of integrating multiple mutation modes for signature discovery and patient stratification, and provides a statistical modeling framework to incorporate additional features of interest for future studies.

Mechanisms underlying genome instability mediated by formation of foldback inversions in Saccharomyces cerevisiae

“…W. Zhang et al, 2018). Moreover, the computationally determined structural variation signature for these cancer-associated foldback inversions contains clustered inverted duplications and deletions (Funnell et al, 2019), which are consistent with the structure of the GCRs observed here.…”

Novel patterns of complex structural variation revealed across thousands of cancer genome graphs

“…In addition, integration of SV (and/or SNV) derived features using topic models or non-negative matrix factorization may help link linear combinations (i.e. signatures) of complex SV types to environmental exposures and DNA repair defects (Macintyre et al, 2018;Funnell et al, 2019;Alexandrov et al, 2013;Nik-Zainal et al, 2016). Comprehensive characterization of the long-range allelic phase of complex SV across large clinically annotated cohorts, leveraging multi-regional and/or single cell sequencing, will be essential to gain insight into the mutational mechanisms underlying SV evolution.…”

Learning mutational signatures and their multidimensional genomic properties with TensorSignatures

“…Cancer arises through the accumulation of mutations caused by multiple processes that leave behind distinct patterns of mutations on the DNA. A number of studies have analysed cancer genomes to extract such mutational signatures using computational pattern recognition algorithms such as non-negative matrix factorization (NMF) over catalogues of single nucleotide variants (SNVs) and other mutation types [1][2][3][4][5][6][7][8] . So far, mutational signature analysis has provided more than 50 different single base substitution patterns, indicative of a range of endogenous mutational processes, as well as genetically acquired hypermutation and exogenous mutagen exposures 9 .…”

“…Mutational signature analysis via computational pattern recognition draws its strength from detecting recurrent patterns of mutations across catalogues of cancer genomes. As many mutational processes also generate characteristic multi nucleotide variants (MNVs) 10,11 , insertion and deletions (indels) [12][13][14] , and structural variants (SVs) 6,[15][16][17] it appears valuable to jointly deconvolve broader mutational catalogues to further understand the multifaceted nature of mutagenesis.…”