In bachisiozic/mmsig: A Mutational Signature Fitting Tool for Hematological Cancers

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

mmsig

The goal of mmsig is to provide a flexible and easily interpretable mutational signature analysis tool. mmsig was developed for hematological malignancies, but can be extended to any cancer with a well-known mutational signature landscape.

mmsig is based on an expectation maximization algorithm for mutational signature fitting and applies cosine similarities for dynamic error suppression as well as bootstrapping-based confidence intervals and assessment of transcriptional strand bias.

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("evenrus/mmsig")

Example

This is a basic example which shows mmsig usage

library(mmsig)

data(mm_5_col)
data(signature_ref)

setting up the mutational signature reference

# remove canonical AID (SBS84) for genome-wide analysis
# remove the platinum signature (SBS35) because the patients are not platinum exposed

sig_ref <- signature_ref[c("sub", "tri", "SBS1", "SBS2", "SBS5", "SBS8", 
                           "SBS9", "SBS13", "SBS18", "SBS-MM1")] 

head(sig_ref)

Setting up the mutation data

# subset three samples to reduce run time

mm_5_col_subset <- mm_5_col[mm_5_col$sample %in% c("MEL_PD26412a", "MEL_PD26411c", "PD26414a"),]
head(mm_5_col_subset)

Perform mutational signature analysis

# Bootstrapping large datasets with many iterations can significantly increase runtime. 

set.seed(1)

sig_out <- mm_fit_signatures(muts.input=mm_5_col_subset, 
                             sig.input=sig_ref,
                             input.format = "vcf",
                             sample.sigt.profs = NULL, 
                             strandbias = TRUE,
                             bootstrap = TRUE,
                             iterations = 20, # 1000 iterations recommended for stable results
                             refcheck=TRUE,
                             cos_sim_threshold = 0.01,
                             force_include = c("SBS1", "SBS5"),
                             dbg=FALSE)

Plot signature estimates

plot_signatures(sig_out$estimate, 
                samples = T, 
                sig_order = c("SBS1", "SBS2", "SBS13", "SBS5", "SBS8", "SBS9", 
                              "SBS18", "SBS-MM1", "SBS35"))