In mknoll/cnAnalysis450k: Copy Number Analysis for 450k Illumina Methylation Arrays

knitr::opts_chunk$set(echo = TRUE)

Aim

This is a short introduction into CN Analysis with the cnAnalysis450k package. The aim is to identify segmental alterations for a given set of samples. For a complete overview for several ways of analyzing and the neccessary parameters refer to the "completeWorkflow" vignette.

## Load data from the minfiData package or minfiDataEPIC
#dataset <- minfiDataEPIC::RGsetEPIC
dataset <- minfiData::RGsetEx

#Normalize data
normData <- 
minfi:::getCN(minfi::preprocessIllumina(
    dataset))
#Controls
ctrlAll <- normData[,1:3]
ctrlAll[is.infinite(ctrlAll)] <- NA
ctrlAll <- scale(ctrlAll)
ctrl <- apply(ctrlAll, 1, "median")
#Samples
samples <- normData[,4:6]
samples[is.infinite(samples)] <- NA
samples <- scale(samples)

Calculate and display Segment-Alterations

##Process data
## ~15 sek
candidatesDATA <-
    cnAnalysis450k::findSegments(samples[, , drop = FALSE], ctrl, ctrlAll)
candidatesMATRIX <-
    cnAnalysis450k::createSegmentMatrix(candidatesDATA, p.select = 0.01)
candidatesCUT <-
    cnAnalysis450k::findCutoffs(candidatesMATRIX, proximity = c(2, 1))
candidatesFINAL <-
    cnAnalysis450k::segmentData(candidatesMATRIX, candidatesCUT, effectsize =
                                    c(0.15, 0.1))


##Display
anno_row <-
    data.frame(do.call(rbind, strsplit(rownames(candidatesFINAL), ":")))
rownames(anno_row) <- rownames(candidatesFINAL)
colnames(anno_row) <- c("Chromosome", "Startpos")
anno_row$Chromosome <-
    factor(anno_row$Chromosome, levels = paste("chr", 1:22, sep = ""))
pheatmap::pheatmap(candidatesFINAL,
                   annotation_row = anno_row[, "Chromosome", drop = FALSE],
                   cluster_rows = FALSE)

Calculate Bin-Alterations

##Process data
candidatesDATA <-
    cnAnalysis450k::createBinsFast(samples, ctrl,  ctrlAll, binsize = 500000)
candidatesMATRIX <- 
    cnAnalysis450k::createBinMatrix(candidatesDATA, pval=1)
candidatesCUT <-
    cnAnalysis450k::findCutoffs(candidatesMATRIX, ignoreNAs=T)
candidatesFINAL <-
    cnAnalysis450k::segmentData(candidatesMATRIX, candidatesCUT, 
                                effectsize = c(0.05, 0.05))


##Display
anno_row <- data.frame(do.call(rbind, strsplit(rownames(candidatesFINAL), ":")))
colnames(anno_row) <- c("chr", "startCG",  "pos")
rownames(anno_row) <- rownames(candidatesFINAL)
pheatmap::pheatmap(candidatesFINAL[which(anno_row$chr=="chr19"),], annotation_row=anno_row[,"chr",drop=F],
                   cluster_rows = FALSE, cluster_cols=FALSE, show_rownames=F)

Calculate Transcript / Gene Alterations

## All transcripts
tx <-
    GenomicFeatures::transcripts(
        TxDb.Hsapiens.UCSC.hg19.knownGene::TxDb.Hsapiens.UCSC.hg19.knownGene)$tx_name

candidatesDATA <- 
    cnAnalysis450k::getTxValuesFast(samples, ctrl, ctrlAll, tx)
candidatesMATRIX <-
    candidatesDATA$data[which(candidatesDATA$p.val <= 0.05), ]
candidatesCUT <-
    cnAnalysis450k::findCutoffs(candidatesMATRIX, ignoreNAs=T)
candidatesFINAL <-
    cnAnalysis450k::segmentData(candidatesMATRIX, candidatesCUT, effectsize =
                                    c(0.05, 0.05))