computePairedPvals-method: Computes pvalues (Fisher test) on the read counts in this...

In vodkatad/roar: Identify differential APA usage from RNA-seq alignments

Description

This is the third step in the Roar analyses: it applies a Fisher test comparing counts falling on the PRE and POST portion in the treatment and control conditions for every gene. The paired method should be used when the experimental setup offers multiple paired samples for the two conditions: that is foreach sample of the control condition there is a naturally paired one for the treatment (i.e. cells derived from the same plate divided in two groups and treated or not). For example in the below code sample treatment sample n.1 (rd1) is paired with control n.2 (rd4) and rd2 with rd3. The pvalue resulting from Fisher test applied on the different samples pairings will be combined with the Fisher method, therefore the pairs of samples should be independent between each other.

Usage

      computePairedPvals(rds, treatmentSamples, controlSamples)
     

Arguments

rds	The RoarDataset or the RoarDatasetMultipleAPA which contains the counts over PRE-POST portions in the two conditions to be compared via pvalues.
treatmentSamples	Numbers that represent the indexes of the treatmentBams/GappedAlign parameter given to the RoarDataset costructor and the order in which they are paired with control samples.
controlSamples	Numbers that represent the indexes of the controlBams/GappedAlign parameter given to the RoarDataset costructor and the order in which they are paired with treatment samples.

Value

The RoarDataset or the RoarDatasetMultipleAPA object given as rds with the compute pvalues phase of the analysis done. Pvalues will be held in the RoarDataset object itself in the case of single samples, while in a separate slot otherwise, but end user normally should not analyze those directly but use totalResults or fpkmResults at the end of the analysis.

Examples

   library(GenomicAlignments)
   gene_id <- c("A_PRE", "A_POST", "B_PRE", "B_POST")
   features <- GRanges(
      seqnames = Rle(c("chr1", "chr1", "chr2", "chr2")),
      strand = strand(rep("+", length(gene_id))),
      ranges = IRanges(
         start=c(1000, 2000, 3000, 3600),
         width=c(1000, 900, 600, 300)),
      DataFrame(gene_id)
   )
   rd1 <- GAlignments("a", seqnames = Rle("chr1"), pos = as.integer(1000), cigar = "300M", strand = strand("+"))
   rd2 <- GAlignments("a", seqnames = Rle("chr1"), pos = as.integer(2000), cigar = "300M", strand = strand("+"))
   rd3 <- GAlignments("a", seqnames = Rle("chr2"), pos = as.integer(3000), cigar = "300M", strand = strand("+"))
   rd4 <- GAlignments("a", seqnames = Rle("chr2"), pos = as.integer(3400), cigar = "300M", strand = strand("+"))
   rds <- RoarDataset(list(rd1,rd2), list(rd3, rd4), features)
   rds <- countPrePost(rds, FALSE)
   rds <- computeRoars(rds)
   rds <- computePairedPvals(rds, c(1,2), c(2,1))
    

vodkatad/roar documentation built on March 30, 2020, 2:56 p.m.