panGeneSets: Test gene networks for a set of cases versus a set of...
In sinnweja/panoply: Personalized Cancer Report

Description Usage Arguments Details Value Examples

Test differential gene expression in gene networks for a set of cases versus a set of controls, with gene networks defined by reactome directed graph for cancer genes, and genomic events determined from variant, cna, or gcount.

panGeneSets(caseids, controlids, variant = NULL, cna = NULL,
           gcount, tumorpct = 0.5, tailPct = 0.1, tailEnd = "both",
           eventOnly = FALSE,  gene.adj = NULL, drug.adj = NULL,
           gageCompare = ifelse(length(caseids) > 1, "as.group", "unpaired"))
drivDNA(ids, variant = NULL, cna = NULL, tumorpct = 0.5, gene.adj=NULL)
outRNA(ids, gcount, tailPct = 0.1, tailEnd = "both")

`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
`ids`	identifiers of subjects in row names of variant and column names of gcount and cna matrices
`variant`	data.frame for somatic and/or germline variants. Must contain columns CHROM, POS, Gene.Symbol, SampleType (Germline, Tumor) and PatientID that must match the caseid and controlid
`cna`	data.frame with log2 of copy number aberrations. Required columns CHROM, START, STOP, Gene.Symbol, and columns named to match caseid and controlid
`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.
`tumorpct`	Vector of approximate tumor percentage for each patient samples given in the order they appear in cases and controls. If a single value is passed, it is used for all patients. Copy number mutation events are called based on their approximate tumor percentage, transformed to the log2-ratio copy number call.
`tailPct`	percentile used to define the most extreme RNA expression events as cancer events per patient; only used when eventOnly=TRUE. Values 0 < tailPct < .5 allowed.
`tailEnd`	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.
`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).
`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.
`gageCompare`	Character string telling the gage package, which performs the gene network tests, to perform tests of 1-vs-M ("unpaired") or N-vs-M ("as.group")
`eventOnly`	Logical, test only gene networks that have a cancer event (variant, cna, and rna-expression outside tailPct)).

If eventOnly=FALSE, then test all gene networks for differential expression between cases and controls. Otherwise, cancer events are determined per gene from variant (any variant in the gene for the case(s)), cna (copy gain or loss for case(s) given tumorpct), and gcount (outlying expression from tailPct and tailEnd settings) and only those gene networks with a cancer event are tested.

A data.frame with the following columns

Cancer.Gene: genes whose networks were tested
Network: total connections of the cacer gene to other genes via reactome adjacency matrix
Network.pval: p-value comparing gene expression between caseids and controlids patients
Network.mean: Z-statistic of differential expression between caseids and controlids
Druggable: Is the cancer gene druggable? 1=TRUE
Total.Druggable: Number of the connected network genes that are druggable
NetGenes: Network genes that are over- or under-expressed
NetDrugs: Drug names that would target the genes in Total.Druggable
Drugs: Drug names that would target the cancer gene

data(clinPanTNBC)
data(genelistPan)
data(cnaPanTNBC)
data(gcPanTNBC)
data(gcinfoPan)
data(variantPanTNBC)
data(dgidbPan)
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)
drivGenes[1:10,1:6]