Computes TF-modules pathway activities scores

This function computes pathway activity scores from normalized gene expression data using a multivariate linear model (MLM) based on the PROGENy resource (Schubert et al., 2018). Optionally, it also performs Gene Set Variation Analysis (GSVA) using hallmark signatures or any user-provided gene sets.

Usage

compute.pathway.activity(
  RNA.tpm,
  gene_sets = NULL,
  paths = NULL,
  return = TRUE,
  file.name = NULL
)

Arguments

RNA.tpm

A numeric matrix of normalized gene expression values with genes as rows and samples as columns.

gene_sets

A named list of gene sets (e.g., hallmark signatures or user-defined sets). If provided, GSVA scores will be computed for these sets. Default is NULL.

paths

A data frame describing the pathway-gene interactions (columns source, target, weight) for use with PROGENy. If NULL, the human PROGENy resource (top 500 genes) is used and cached in Results/Pathways_collection_PROGENy.csv.

return

Logical; if TRUE, saves matrices in Results/ folder. Default is TRUE.

file.name

Optional character suffix used when writing output CSV files.

Value

If gene_sets is NULL, a scaled data frame of PROGENy pathway activity scores (samples as rows, pathways as columns). If gene_sets is provided, a list with two elements:

• PROGENy: A scaled data frame of PROGENy pathway activity scores.

• GSVA: A scaled data frame of GSVA scores based on the provided gene sets.

Column names are made syntactically valid with make.names().

References

Schubert M, Klinger B, Kluenemann M, Sieber A, Uhlitz F, Sauer S, Garnett MJ, Bluethgen N, Saez-Rodriguez J. Perturbation-response genes reveal signaling footprints in cancer gene expression. Nature Communications. 2018. doi:10.1038/s41467-017-02391-6

Examples

if (FALSE) { # \dontrun{
# Compute only PROGENy activities (downloads the PROGENy model from OmniPath)
data("counts.norm.tuto")
pathways <- compute.pathway.activity(counts.norm.tuto)
} # }