fpkmResults-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).

1 2	fpkmResults(rds)

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.

The resulting dataframe will be identical to that returned by totalResults but with two columns added: "treatmentValue" and "controlValue". These columns will contain a number that indicates the level of expression of the relative gene in the treatment (or control) condition. For RoarDataset this number derives from the counts (averages across samples when applicable) obtained for the PRE portion of the gene and is similar to the RPKM standard measure of expression used in RNAseq experiment. Specifically we correct the counts on the PRE portions dividing them by portion length and total numer of reads aligned on all PRE portions and the multiply the results for 1000000000. See the vignette for more details.

For RoarDatasetMultipleAPA the same procedure is applied to all the possible PRE choices for genes. Note that summing all the counts for every PRE portion assigned to a gene could lead to count some reads multiple times when summing all the PRE portions counts therefore this measure is not completely comparable with the one obtained with the single APA analysis. The length column added in this case contains the length of the PRE portions (counting only exonic bases).

   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 <- fpkmResults(rds)