Hey there,
It seems to me that due to how most RNAseq libraries are now constructed, most people who want to count reads per feature with summarizeOverlaps will actually want to use ignore.strand = FALSE and preprocess.reads = invertStrand. However, I only stumbled across this after realizing that my output was wildly different from what htseq-count with -s reverse gave me when I only used ignore.strand = FALSE.
That you need to use preprocess.reads = invertStrand with most modern stranded RNAseq libraries is not mentioned in the documentation for summarizeOverlaps or in the example RNAseq workflow (which is not this package, but it's still confusing). The only information that I could find to lead me to this was in the bioconductor support forums.
Could the documentation be clarified in some way? I could see this tripping up new users.
Hey there,
It seems to me that due to how most RNAseq libraries are now constructed, most people who want to count reads per feature with summarizeOverlaps will actually want to use
ignore.strand = FALSEandpreprocess.reads = invertStrand. However, I only stumbled across this after realizing that my output was wildly different from what htseq-count with-s reversegave me when I only usedignore.strand = FALSE.That you need to use
preprocess.reads = invertStrandwith most modern stranded RNAseq libraries is not mentioned in the documentation for summarizeOverlaps or in the example RNAseq workflow (which is not this package, but it's still confusing). The only information that I could find to lead me to this was in the bioconductor support forums.Could the documentation be clarified in some way? I could see this tripping up new users.