featureCounts strandedness option

Question

[Noah]

Hi all,

I am using featureCounts to quantify my RNA-seq signal over gene regions. I am using it as follows:

For my single-end reads: featureCounts -t exon -g gene_id -s 0 -O -T 10 -a <.gtf> -o <.counts> <.bam>

For my paired-end reads: featureCounts -p -t exon -g gene_id -s 0 -O -T 10 -a <.gtf> -o <.counts> <.bam>

Where I am running into something confusing is with the -s option. I have run these with all options for -s, with differing proportions of 'successfully assigned reads'. -s 0 (unstranded) and -s 2 both assign about ~25% of the reads, however, -s 1 assigns only ~3% of reads to gene regions. The fact that both 0 and 2 are producing similar proportions of assigned reads/fragments is interesting, and must say something about the library... but I need help determining exactly what that is.

Thanks in advance. Let me know if any additional information is needed.

[Istvan Albert]

Another important detail when counting paired end reads is that the meaning of -p has changed:

Release 2.0.2, 29 March 2021

New parameter --countReadPairs is added to featureCounts to explicitly specify that read pairs will be counted, and the -p option in featureCounts now only specifies if the input reads are paired end (it also implied that counting of read pairs would be performed in previous versions).

basically when read pairs are counted the counts will be halved

[rpolicastro]

Your library is reverse stranded, so if you set it to stranded (-s 1) you would be aligning most of your reads to the wrong strand relative to the gene coding direction. With unstranded (-s 0) you align to both strands so you would capture any transcripts from that section of DNA regardless if the library was unstranded, stranded,or reverse stranded.

Answer 1

[Istvan Albert]

-s 2 captures sense transcription,

-s 1 captures the antisense transcription.

-s 0 captures both

so it seems you have low-levels of antisense transcription,

everything is working as expected