About

This package carries out Network-based Gene Set Analysis by incorporating external information about interactions among genes, as well as novel interactions learned from data.

How to install?

You can install it directly from CRAN:

install.packages("netgsa",build_vignettes=T)

Reference manual is available on CRAN. A vignette on how to use NetGSA is also available. A development version is available on GitHub and can be installed via the following:

library(devtools)
devtools::install_github("drjingma/netgsa", build_vignettes=T)

More details about the method implemented can be found in the original paper here and a follow-up review paper here.

Why should someone use netgsa?

• NetGSA incorporates the rich network information curated in public databases (e.g. KEGG, reactome, etc.) and/or learned from high-throughput sequencing data, thereby gaining power in detecting active genetic/metabolic pathways.

How does it compare to other methods?

• NetGSA tests the self-contained null hypothesis and compares the set of genes in a given pathway with itself.
• NetGSA allows users to complement potentially misspecified information in public databases with high-throughput sequencing data.
• NetGSA is particularly powerful in detecting active metabolic pathways from metabolomic data where the pathway size is relatively small and available metabolic network information is sparse.
• See more details in our review paper here.

Notes

• NetGSA is based on a linear mixed model. Estimation of the variance components in this model can be done via restricted maximum likelihood or via the restricted Haseman-Elston (REHE) regression. See our recent paper here for fast variance components estimation with REHE.

• NetGSA is seamlessly integrated with external databases on gene-gene interactions and utilizes Cytoscape for interactive visualization of enriched pathways. Moreover, NetGSA can handle thousands of genes within minutes. See here for details on how we improved the computation and visualization of NetGSA.