cn_calcPRs: assess performance of GRN predictions based on zscores

In pcahan1/CellNet: Assess and improve cell fate engineering with network biology

assess performance of GRN predictions based on zscores

Description

Recall (Sensitivity): fraction of gold standard regulatory relationships detected by GRN. Precision (1-FPR): Proportion of called relationships that are in the gold standard

Usage

cn_calcPRs(zscores, tfs, goldStandard, universe = NULL, tRange = seq(3,
  6, by = 0.25), random = FALSE, randomIter = 10,
  funType = "intersect")

Arguments

zscores	matrix of zscores
tfs	vector of tf(s) perturbed
goldStandard	gold stanrd gene list
universe	genes tested
tRange	threshold sequence
random	boolean, whether to compute a random selection of calls, too, based on the number of genes actually predicted
randomIter	numer of random selects to make, returns only the average
funType	how to treat >1 TF, intersect or union

Value

List of performance results of form: data frame of Score (cutoff), Precision, Recall, Predictions (n), obs = real GRN, rand = list of random GRNs

Examples

grnPR<-cn_calcPRs(zscs, "Sox2", aGoldStandard[["Sox2"]], allGenes, tRange=seq(0,8, by=.5), random=TRUE, randomIter=3);

pcahan1/CellNet documentation built on May 18, 2023, 4:58 p.m.