Scoring of Drugs via Network-based Methods

Usage

score_drugs(driveR_res, drug_interactions_df, W_mat, method, ...)

Arguments

driveR_res: data frame of driveR results
drug_interactions_df: data frame of drug-gene interactions
W_mat: adjacency matrix for the PIN
method: scoring method (one of 'distance-based' or 'RWR')
...: additional arguments for score_drugs_distance_based or score_drugs_RWR_based

Value

vector of scores per drug.

Details

This is the wrapper function for the two proposed methods for personalized scoring of drugs for individual cancer samples via network-based methods. The available methods are 'distance-based' and 'RWR'. For the 'distance-based' method, the score between a gene (g) and drug (d) is formulated as: $$score(g, d) = driver(g) / (d(g, d) + 1)^2$$ where driver(g) is the driverness probability of gene g, as predicted by 'driveR' and d(g, d) is the distance withing the PIN between gene g and drug d. The final score of the drug d is then the average of the scores between each altered gene and d: $$score(d) = \Sigma{score(g, d)} / |genes|$$

For the 'RWR' method, a random-walk with restart framework is used to propagate the driverness probabilities.

By default DGIdb_interactions_df is used as the drug_interactions_df.

If the W_mat argument is not supplied, the built-in STRNG data STRING_adj_df is used to generate W_mat.

Examples

toy_data <- data.frame(
  gene_symbol = c("TP53", "EGFR", "KDR", "ATM"),
  driverness_prob = c(0.94, 0.92, 0.84, 0.72)
)
toy_interactions <- DGIdb_interactions_df[1:25, ]
res <- score_drugs(
  driveR_res = toy_data,
  drug_interactions_df = toy_interactions, # leave blank for default
  W_mat = toy_W_mat, # leave blank for default
  method = "distance-based",
  verbose = FALSE
)
#>