fastGeneMI: an R package for computing the mutual information between the expression profiles of genes in a microarray
The following R packacges are required (available on CRAN unless otherwise stated):
- Rcpp
- RcppArmadillo
- infotheo
- PRROC
- minet (Bioconductor)
- testthat
Requires devtools. Install from R using
devtools::install_github("jonathanishhorowicz/fastGeneMI").
fastGeneMI is a package that implements a suite of mutual information estimators in parallelised C++ with an R interface.
These were implemented for inferring gene regulatory networks from DNA microarrays but are suitable for any continuous data.
The following estimators are implemented:
- Maximum Likelihood
- Chao-Shen
- Miller-Madow
- Shrinkage
- B-spline
Also included are the inference algorithms CLR, MRNET and ARACNe, as well as functions to evaluate their predictions using precision-recall curves. Included data is from the DREAM5 Network Inference Challenge. See the Supplementary Material of Ish-Horowicz and Reid, 2017 for more detailed descriptions of each estimator.
Full documentation is available in fastGeneMI-manual.pdf.