Identify Mutually Exclusive Mutations.
Introduction to Rediscover
We present Rediscover, an R package to identify mutually exclusive genomic events. It reimplements a privious R package (Discover) whose main contribution is a statistical analysis based on the Poisson-Binomial distribution that takes into account that some samples are more mutated than others. Rediscover is much faster than the discover implementation.
Installation
Rediscover can be installed from CRAN repository:
install.packages("Rediscover")
Introduction
The package library has two main parts:
- Estimation of the probabilities $p_ {ij}$ that gene i is mutated in sample j -assuming conditional independence between genes and samples-.
- Estimation of p-values using the Poisson-Binomial distribution, using the previous probabilities and the number of samples in which two genes are co-mutated. The corresponding null hypothesis $H_0$ is that the mutational status of both genes is independent of each other.
The second step is the estimation of the p-values using these probabilities and the number of samples where two genes are co-altered. Rediscover offers different functions depending on the desired study:
getMutex
if the user wants to evaluate if genes are mutually exclusive.getMutexAB
if the user wants to evaluate if genes are mutually exclusive with respect to another event (amplifications, deletions, etc...)getMutexGroup
will be used when the user wants to obtain the probability that a certain group of genes being mutually exclusive. Unlike thegetMutex
function, in this case the users introduces the set of genes of interest.
Rediscover also provides a function to integrate its usage with maftools
and TCGAbiolinks
. Specifically, we added to the function somaticInteractions
from maftools
our analyses based on the Poisson-Binomial distribution resulting in a new function called discoversomaticInteractions
.