panDrugSets: Tests on drugs for their effectiveness in targeting cancer...
In sinnweja/panoply: Personalized Cancer Report

Description Usage Arguments Details Value Examples

Tests on drugs for their effectiveness in targeting activated cancer genes in cases against their matched controls.

panDrugSets(panGene, caseids, controlids, gcount, minTargets=1, minPathPct=.05, minPathSize=8,
            minPathways=1, nsim=1000, tailEnd="both", gene.gs=NULL,
            gene.adj=NULL, drug.gs=NULL, drug.adj=NULL, 
            gageCompare = ifelse(length(caseids) > 1, "as.group", "unpaired"))

`panGene`	A data.frame of drug test results from panGeneSets
`caseids`	identifiers of subjects case subjects that will match to variant column PatientID, and column name of cna and gcount.
`controlids`	identifiers of subjects in row names of variant and column names of gcount who are controls
`gcount`	Normalized gene expression counts for patients (columns) at gene symbols (rows). Gene symbols are the row names and should match dataset gcinfoPan gene symbols.
`minTargets`	Retain drug results for drugs that target at least minTargets genes in the cancer and network genes.
`minPathPct`	Of reactome pathways affected by gene networks targeted by the drug, only count those with total druggable by the drug in the pathway divided by the total pathway size >= minPathPct.
`minPathSize`	Of reactome pathways affected by gene networks targeted by the drug, only count the pathways that have total genes >= minPathSize.
`minPathways`	Filter drug results to those that target genes in at least minPathways, after accounting for minPathPct and minPathSize.

gene.gs=NULL, gene.adj=NULL, drug.gs=NULL, drug.adj=NULL,

`drug.gs`	A list of genes, where the genes in each set are targeted by each drug in drug.adj.
`drug.adj`	Adjacency matrix (values of 1 or 0) for drug-gene interactions with the drug in the row directly impacting the gene in the column. The framework is equipped for coding for other interaction types of the drug and genes.
`gene.gs`	List of genes per pathway/network to annotate the pathways targeted by the drug. Can be user-defined, but panoply contains reactome sets.
`gene.adj`	Adjacency matrix of genes connected via a graph of nodes and edges. Panoply contains reactome.adj, which is directed acyclic graph of Reactome nodes (genes) and edges (relationships).
`tailEnd`	For Drug Network tests, perform differential expression tests that are one-sided ("upper" or "lower"), or two-sided ("both"). Also used with tailPct to include both, upper, or lower expression events to be used as cancer events.
`nsim`	Number of simulations to perform for evalulating significance of meta drug tests
`gageCompare`	Character string for the R gage package to specify how to do within-network comparisons of the case(s) versus the controls across genes

If tailEnd is upper, then testing for drugs that target over-expressed genes and gene networks in case(s) versus controls.

A data.frame with the following columns

Drug: Drug Name
N.Cancer.Genes: number of cancer genes that are targeted by the drug
Cancer.Genes: cancer genes that are targeted by the drug
N.Network.Genes: number of network genes targeted by the drug
Network.Genes: network genes targeted by the drug
N.Pathways: number of pathways with genes targeted by the drug (meeting minPathPct and minPathSize settings)
Pathways: pathway names of pathways with a gene targeted by the drug
Network: Number of genes directly targeted by the drug
DNT.pval: p-value comparing gene expression between caseids and controlids patients in the genes directly targeted by the drug
DMT.Stat: Meta Z-statistic of combined gene networks for any gene targeted by the drug
DMT.pval: p-value for DMT.Stat Z-statistic, based on multiplying s_i~N(0,1) by gene network test Z-statistics for i=1..nsim
Z.stat: Meta Z-statistic of combined gene networks for any gene targeted by the drug, weighted by number of genes targeted by the drug divided by gene network size
PScore: A score used to sort the drug test results, which is a sum of the -log10(pval) of Network.pval and ZSim.pval

data(clinPanTNBC)
data(genelistPan)
data(cnaPanTNBC)
data(gcPanTNBC)
data(gcinfoPan)
data(variantPanTNBC)
data(dgiSets)
data(reactome)
patient <- "TCGA-B6-A0IK"
match.index <- which(clinPanTNBC$Vital.Status=="Alive" & clinPanTNBC$PatientID!=patient & clinPanTNBC$age_at_diagnosis>50 & clinPanTNBC$N.stage == "N1" & clinPanTNBC$ndays > 1000)
ptmatch <- clinPanTNBC[match.index,"PatientID"]
drivGenes <- panGeneSets(caseid=patient, controlid=ptmatch, eventOnly=TRUE, variant=variantPanTNBC,
  cna=cnaPanTNBC, gcount=gcPanTNBC, tumorpct=0.3, tailEnd="upper", tailPct=0.1)
set.seed(1000)
drugResults <- panDrugSets(drivGenes, caseids=patient,
controlids=ptmatch, gcount=gcPanTNBC,nsim=200, tailEnd="upper",
drug.gs=dgi.gs, drug.adj=dgi.adj, gene.gs=reactome.gs, gene.adj=reactome.adj)

drugResults[1:10,!grepl("^Pathway", names(drugResults))]