pmsignature

The R package pmsignature is developed for efficiently extracting characteristic mutation patterns (mutation signatures) from the set of mutations collected typically from cancer genome sequencing data.

For extracting mutation signatures, principal component analysis or nonnegative matrix factorization have been popular. Compared to these existing approaches, the pmsignature has following advantages:

pmsignature can perform robust estimation of mutation signatures in case of many contextual factors are taken into account such as two bases 5' and 3' to the mutated sites.
pmsignature provides intuitively interetable visualization of mutation signatures, which is analogous to sequencing logos.

Paper

Shiraishi et al. A simple model-based approach to inferring and visualizing cancer mutation signatures, bioRxiv, doi: http://dx.doi.org/10.1101/019901.

Input data

Here, mutation features are elements used for categorizing mutations such as:

6 substitutions (C>A, C>G, C>T, T>A, T>C and T>G)
5’ and 3’ flanking bases (A, C, G and T)
transcription direction.

Currently, pmsignature can accept following two formats of tab-delimited text file.

Mutation Position Format

sample1 chr1 100 A C sample1 chr1 200 A T sample1 chr2 100 G T sample2 chr1 300 T C sample3 chr3 400 T C

The 1st column shows the name of samples
The 2nd column shows the name of chromosome
The 3rd column shows the coordinate in the chromosome
The 4th column shows the reference base (A, C, G, or T).
The 5th colum shows the alternate base (A, C, G, or T).

Mutation Feature Vector Format

1 4 4 4 3 3 2 2 4 3 3 1 1 2 3 4 4 3 2 2 2 4 3 3 2 3 3 1 5 3 4 2 4 4 2 6 4 1 4 2 1 2 3 2 1 1 1 1 2 7 4 2 2 4 3 2

The 1st column shows the name of samples
From the 2nd to the last column show the value of mutation features.
In this example, substitution patterns (1 to 6 values, C>A, C>G, C>T, T>A, T>C and T>G), two 5' and 3' flanking bases (1 to 4 values for each 4 site, A, C, G and T) and transcription direction (1 to 2 values, + and -) are considered.
However, you can put your favorite mutation features in this file format.

Workflow

Install the package

First, several R packages such as ggplot2, Rcpp, GenomicRanges, BSgenome.Hsapiens.UCSC.hg19, which pmsignature depends has to be installed. Also, devtools may be necessary for ease of installation. Since the pmsignature utilizes C++ codes by way of Rcpp, you need to install C++ compiler (e.g., Rtools for Windows, Xcode for Mac). See Advanced R by Dr. Hadley Wickham about the basic usage of Rcpp.

source("http://bioconductor.org/biocLite.R")
biocLite(c("GenomicRanges", "BSgenome.Hsapiens.UCSC.hg19"))
install.packages("devtools")
install.packages("ggplot2")
install.packages("Rcpp")

Currently, the easiest way for installing pmsignature is to use the package devtools:

library(devtools)
devtools::install_github("friend1ws/pmsignature")
library(pmsignature)

For those who failed to installing, we recommend to use the newest R version. Also, you may be required to upgrade the bioconductor:

source("http://bioconductor.org/biocLite.R")
biocLite("BiocUpgrade")

Prepare input data

First, create the input data from your mutation data.

After installing pmsignature, you can find example files at the directory where pmsignature is installed:

Mutation Position Format

inputFile <- system.file("extdata/Nik_Zainal_2012.mutationPositionFormat.txt.gz", package="pmsignature")
print(inputFile)

Mutation Feature Vector Format

inputFile <- system.file("extdata/Hoang_MFVF.ind.txt.gz", package="pmsignature")
print(inputFile)

Read input data

Type the following commands:

Mutation Position Format

G <- readMPFile(inputFile, numBases = 5)

Here, inputFile is the path for the input file. numBases is the number of flanking bases to consider including the central base (if you want to consider two 5' and 3' bases, then set 5). You can format the data as the full model by typing

G <- readMPFile(inputFile, numBases = 5, type = "full")

Also, you can add transcription direction information by typing (in that case, the package TxDb.Hsapiens.UCSC.hg19.knownGene is necessary)

G <- readMPFile(inputFile, numBases = 5, trDir = TRUE)

Mutation Feature Vector Format

G <- readMFVFile(inputFile, numBases = 5, type="independent", trDir=TRUE)

Estimate the parameters

When you want to set the number of mutation signature as 3, type the following command (see also ?getPMSignature):

Param <- getPMSignature(G, K = 3)

If you want to add the background signature, then after obtaining the background probability, perform the estimation. Currently, we only provide the background data for the "independent" and "full" model with 3, 5, 7 and 9 flanking bases (see also ?readBGFile).

BG_prob <- readBGFile(G)
Param <- getPMSignature(G, K = 3, BG = BG_prob)

In default, we repeat the estimation 10 times by changing the initial value, and select the parameter with the highest value of log-likelihood. If you want to changet the trial number, then

Param <- getPMSignature(G, K = 3, numInit=20)

Visualing the mutation signatures and memberships

To visualize the mutation signatures by typing (see also ?visPMSignature):

visPMSignature(Param, 1)
visPMSignature(Param, 2)
visPMSignature(Param, 3)

To obtain the value of estimated mutation signatures (see also ?getSignatureValue):

getSignatureValue(Param, 1)
getSignatureValue(Param, 2)
getSignatureValue(Param, 3)

To see the overview of the estimated membership parameter (see also ?visMembership):

visMembership(G, Param)

Here, samples are sorted according to the number of mutations. To unsort the sample, set sortSampleNum = FALSE. Also, not to multiply the number of mutations to the barplot, set multiplySampleNum = FALSE.

To obtain the value of estimated membership parameters (see also ?getMembershipValue):

getMembershipValue(Param)

diegopenhanut/pmsignature