-s=reverse HTSEQ option

[lwhitmore]

Hi,
When using the -s=reverse option for HTSEQ most reads counted are mapped to anti-sense strand of the gene. How does this option work and what exactly is it doing?

Thanks,
Leanne

[apredeus]

some library-prep methods result in "strand reversal" - basically your cDNA reads map to the strand opposite to where actual gene is. So if you're not sure what library prep was used, run Picard's CollectRNAseqMetrics on your experiment, and see what percent of the mapped reads matches that of the "feature" (that is, annotated gene). If you get something like 50%, use unstranded option. If you get 100%, use stranded. And if you get about 0%, use reverse.

This holds true not only for htseq but also for RSEM and all other strand-aware Rna-Seq quantifiers

[lwhitmore]

right so if we know that we have strand reversed library and we use the option -s reverse then reads will only be counted if they map to the apposing strand that the feature "gene" is actually on correct?

Thanks
Leanne

[apredeus]

Well if you are using -s reverse, the reads landing on the transcript in the same direction as the transcript should NOT be counted. Really, what you should do is run quantification on your library with all three options and see what sort of numbers you get. Say, for example, if you had 40M aligned reads, of which 32M land on the annotated transcripts, you should get the sum total of all genes as follows:

- s no = 32 M
- s reverse = 30-31 M
- s yes = 1-2 M

if that's not what you see, you need to dig into your data and see what went wrong.