pvalueCorrectFilter-method: Returns a dataframe with results of the analysis for a...
In vodkatad/roar: Identify differential APA usage from RNA-seq alignments

Description Usage Arguments Value Examples

The last step of a classical Roar analyses: it returns a dataframe containing m/M values, roar values, pvalues and estimates of expression (a measure recalling FPKM). Only the genes with an expression estimate bigger than a given cutoff will be considered. Also pvalues, corrected considering multiple testing, will be considered for filtering.

1 2	pvalueCorrectFilter(rds, fpkmCutoff, pvalCutoff, method)

`rds`	The `RoarDataset` or the `RoarDatasetMultipleAPA` with all the analysis steps (`countPrePost`, `computeRoars`, `computePvals`) performed. If one or more steps hadn't been performed they will be called automatically.
`fpkmCutoff`	The cutoff that will be used to determine if a gene is expressed or not.
`pvalCutoff`	The cutoff that will be used to determine if a pvalue is significative or not.
`method`	The multiple test correction method that has to be used (used only for multiple paired samples or single samples, not used for multiple unpaired samples.)

For RoarDataset:

The resulting dataframe will be identical to that returned by standardFilter but after gene expression filtering another step will be performed: for single samples comparisons or multiple paired samples comparisons only genes with a corrected (with the given method) pvalue (for paired datasets this is the combined pvalue obtained with the Fisher method) smaller than the given cutoff will be returned, while for multiple samples a column (nUnderCutoff) will be added to the dataframe. This column will contain an integer number representing the number of comparisons between the samples of the two conditions that results in a nominal pvalue lower than the given cutoff (pvalCutoff).

For RoarDatasetMultipleAPA: for each gene we select the APA choice that is associated with the smallest p-value then proceed exactly as above.

   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("+"))
   rds <- RoarDataset(list(c(rd1,rd2)), list(rd3), features)
   rds <- countPrePost(rds, FALSE)
   rds <- computeRoars(rds)
   rds <- computePvals(rds)
   dat <- pvalueFilter(rds, 1, 0.05)