Hisat R strandness aligning same strand as F strandness, how to do it correctly?

Question

[Niek De Klein]

I have RNA-sequence data where the RNA was isolated using the ribo-zero protocol. I (presumably) have the reverse complement of the transcript. I aligned them using Hisat2 with

hisat2 -x index -U input.fastq -S output.sam --rna-strandness R

However, when viewing it in IGV the reads are always aligned to the wrong strand of the annotated genes. For example in the figure IL18RAP is on the + strand, but the reads get aligned to the - strand (top sample, blue reads are - strand with <- direction, red reads are + strand with -> direction. IL18RAP has -> direction).

I tried doing the same but aligning to forward strand as I thought maybe I don't have the reverse complement of the transcript

hisat2 -x index -U input.fastq -S output.sam --rna-strandness F

but this also aligns to the wrong strand (middle sample).

Finally, I took the reverse complement of the reads and did the alignment with --rna-strandness F, and now they did align to the correct strand. However, I thought that this should be the same with --rna-strandness R, as the documentation says 'R' means a read corresponds to the reverse complemented counterpart of a transcript. How can I correctly allign to the reverse complement strand with Hisat?

Answer 1

[WouterDeCoster]

I also noticed that the --rna-strandness parameter is essentially doing nothing. I have emailed the authors of the tool this issue but never received a reply.

[prasundutta87]

Hi,

Did you get a reply from HISAT2 people regarding --rna-strandness? I have actually used HISAT2 wth only --dta (Report alignments tailored for transcript assemblers ) for paired end stranded RNAseq reads (Truseq). The generated SAM/BAM files have been used for differential expression analysis. But, I am also using them for measuring allele specific expression. Not completely sure if --rna-strandness matter in this case?

Answer 2

[agolicz]

It looks to me like Hisat2 does not reverse complement the read but adds correct XS tag. Which is pretty much what it says in the manual. If you mouse over the read you will see that the XS tag (either + or -) is correct with respect to annotated transcript.