mSigHdp: hierarchical Dirichlet process mixture modeling for mutational signature discovery

Abstract

Mutational signatures are characteristic patterns of mutations caused by endogenous or exogenous mutational processes. These signatures can be discovered by analyzing mutations in large sets of samples - usually somatic mutations in tumor samples. Most programs for discovering mutational signatures are based on non-negative matrix factorization (NMF). Alternatively, signatures can be discovered using hierarchical Dirichlet process (HDP) mixture models, an approach that has been explored less. These models assign mutations to clusters and view each cluster as being generated from the signature of a particular mutational process. Here we describe mSigHdp, an improved approach to using HDP mixture models to discover mutational signatures. We benchmarked mSigHdp and state-of-the-art NMF-based approaches on 4 realistic synthetic data sets. These data sets encompassed 18 cancer types. In total they contained 3.5x107 single-base-substitution mutations representing 32 signatures and 6.1x106 small-insertion-and-deletion mutations representing 13 signatures. For 3 of the 4 data sets, mSigHdp had the best positive predictive value for discovering mutational signatures, and for all 4 data sets, it had the best true positive rate. Its CPU usage was similar to that of the NMF-based approaches. Thus, mSigHdp is an important and practical addition to the set of tools available for discovering mutational signatures. Data and code availabilitymSigHdp is available at public repositories https://github.com/steverozen/mSigHdp and https://github.com/steverozen/hdpx. The synthetic data, code for generating the synthetic data, code for running the mutational-signature discovery programs, the main outputs of the programs, and code for analyzing their results and for generating the data figures in this paper are available at https://github.com/Rozen-Lab/mSigHdp sup files. A singularity container with mSigHdp can be downloaded from cloud.sylabs.io with the shell command "singularity pull library://rozen-lab/msighdp/msighdp:2.1.2". A toy-example Rscript for using this container is at https://github.com/steverozen/mSigHdp/blob/master/data-raw/container_scripts/test_mSigHdp.R. Supplementary materialOne excel file of supplementary tables and one PDF file of supplementary figures have been submitted along with this manuscript.