Analyze the the “activities” of mutational signatures in one or more mutational spectra. ‘mSigAct’ stands for mutational Signature Activity. mSigAct uses a maximum likelihood approach to estimate (conservatively) whether there is evidence that a particular set of mutational signatures is present in a spectrum. It can also determine a minimal subset of signatures needed to plausibly reconstruct an observed spectrum. This sparse assign signatures functionality is deliberately biased toward using as few signatures as possible. There is also functionality to do a maximum a posteriori estimate of signature activity, which makes use of information on the proportion of tumors in a given type that have a particular signature combined with the likelihood that a particular combination of signatures generated an observed spectrum.
The concepts behind the signature presence test and the sparse-assign-signature functionality are described in Ng et al., 2017, “Aristolochic acids and their derivatives are widely implicated in liver cancers in Taiwan and throughout Asia”, Science Translational Medicine 2017 https://doi.org/10.1126/scitranslmed.aan6446.
if (!requireNamespace("remotes", quietly = TRUE)) {
install.packages("remotes")
}
remotes::install_github(repo = "steverozen/mSigAct", ref = "v3.0.1-branch")The alpha version used in Ng et al., 2017, “Aristolochic acids and their derivatives are widely implicated in liver cancers in Taiwan and throughout Asia”, Science Translational Medicine 2017 https://doi.org/10.1126/scitranslmed.aan6446 is at https://github.com/steverozen/mSigAct/tree/v1.2-alpha-branch.
To use new features in the development version, you can install mSigAct from the master branch on GitHub, which may not be stable:
if (!requireNamespace("remotes", quietly = TRUE)) {
install.packages("remotes")
}
remotes::install_github(repo = "steverozen/mSigAct", ref = "master")https://github.com/steverozen/mSigAct/blob/v3.0.3-branch/data-raw/mSigAct_3.0.3.pdf