iPAS (integrative Pathway Activity Signatures) is a method for detecting changes in signaling pathway activity via differential expression signatures. Our method uses the LINCS (Library of Integrative Network-Based Cellular Signatures) library of perturbation signatures to construct transcriptomic signatures that can detect alterations in activity of signaling pathways. We integrate transcriptomic response across a panel of 12 diverse cell lines using machine learning methods to create pathway signatures that are strong predictors of pathway activity in a variety of contexts.
install.packages("remotes")
remotes::install_github("NicholasClark/iPAS")See the vignette for an example of usage.
The main function, iPAS_enrich, takes in a matrix of query signatures (column names should be condition/sample names, row names should be entrez gene ids) and outputs an object with z-scores and empirical p-values (computed by a permutation test) for each pathway.
There are functions to view the output as bar charts (iPAS_bar) of top pathways, heatmaps (iPAS_heatmap) of all pathway scores, or a density plot (iPAS_density and iPAS_density_facet) of the z-score for a particular pathway versus the null distribution of permutation scores for that pathway.
Our manuscript is in preparation:
Clark et al., Integrative signatures of signaling pathway response increase robustness and accuracy of pathway predictions
This work builds on previous work published in Bioinformatics (2020):
Ren et al. Predicting mechanism of action of cellular perturbations with pathway activity signatures" https://doi.org/10.1093/bioinformatics/btaa590