R/exprs2fingerprint.R

In pathprint: Pathway fingerprinting for analysis of gene expression arrays

exprs2fingerprint <-
function(exprs, platform, species, progressBar = TRUE)
# Produces a normalized fingerprint directly from an exprs object
# GEOnormalize is used to normalize the score against the GEO background
# The pathways in the fingerprint are a combination of 
# * KEGG
# * Wikipathways
# * KEGG shell
# * Netpath
# * Static networks
#
# Platforms should be of the type listed in GEO (e.g. "GPL570").
# Species can be "mouse" or "human"
# Updated for v0.3
{
    if(class(exprs) == "ExpressionSet"){
            exprs = exprs(exprs)
    }

#######
# Load data and check parameters
#######

# Load annotation and background data associated with each platform   
#    if (!(exists("chipframe"))){
#      print("Loading background file")
#      data(chipframe)
#    }
# check platform is compatible
    if (!(platform %in% names(chipframe))) 
        stop("Platform name is invalid or not currently supported")

# Load pathway gene sets and check species

#    if (!(exists("genesets")))
#    { 
#      data(genesets)
#    }
    if (!(species %in% names(genesets))) 
        stop("Species name invalid or not supported")
#   data(list = genesets[species])
    gsdb <- get(genesets[species]) 

#########
# Run pathway enrichment
#########

    print("Running fingerprint")

# Annotate array according to the annotations
    geo.ann <- customCDFAnn(exprs, chipframe[[platform]]$ann)
    
# Use SCE to calculate a score for each pathway based on the mean or median

SCE <- single.chip.enrichment( 
            exprs = geo.ann,
            geneset = gsdb,
            transformation = "squared.rank",
            statistic = "mean",
            normalizedScore = FALSE,
            progressBar = progressBar
            )

#########
# Threshold according to GEO corpus background 
#########  

    SCE.threshold<-thresholdFingerprint(SCE = SCE, platform = platform)

    return(SCE.threshold)

# End exprs2fingerprint
}