XS tags in hisat2 alignment

[dovah]

Hi,

I had some troubles running Cufflinks with input mapping from HIsat2 (sam/bam).

Cufflinks man page says that input file should be sorted and contain XS tags. To add the tags, I specified --dta-cufflinks as parameter in for the alignment, following Hisat2 man page. With this parameter, supposedly, " [...] HISAT2 produces an optional field, XS:A:[+-], for every spliced alignment.". However, this tag does not appear in every line (read) of the generated bam file.

Is this possible and/or expectable? If you ever encountered this issue before, how could you fix it?

I was thinking to first generate a file like this prior to launch of Cufflinks run:

samtools view hisat2.sorted.bam | egrep 'XS:A:-|XS:A:+' > XS-Tagged.sorted.sam

What do you think?

[Brian Bushnell]

IIRC, Cufflinks only requires XS tags for spliced alignments.

[dovah]

Ha ok, so it should be fine. Have you ever used Cufflinks on bam file generated by Hisat2 by the way? Because it is taking a huge lot of time. I can spot CPU usage (~99%) but no tmp files in the working directory.

[Brian Bushnell]

Cufflinks ... is taking a huge lot of time.

Sounds like it's working as expected

Seriously, though, sometimes that can happen if you have super-high coverage over certain reasons... in which case filtering out those regions can help (if you're not interested in them).

[dovah]

oh thanks Brian. I was so worried :P