Transcriptome ID mismatch in abundance files

[jynewberg]

When Kallisto is run on a fasta reference with the following header style:

>0 ENST00000456328.2 chr1+ 11869-12227,12613-12721,13221-14409

The feature ID appears as "ENST00000456328.2" in the abundance.tsv output file, but it appears as "0" in the abundance.h5 file.

I noticed this because I was getting

Error in check_target_mapping(tmp_names, target_mapping) :
couldn't solve nonzero intersection

when I ran Sleuth. I'm pretty sure this issue arises because the transcript IDs I provide did not match the abundance.h5 IDs, although they match perfectly with the abundance.tsv IDs.

This isn't a bug, per se, but rather a reporting issue that causes confusion. It would probably be best to have the abundance.tsv IDs match the abundance.h5 IDs, so as to reduce confusion. Note that the header style I am using is what gtf_to_fasta produces when run on GTF files from Gencode or Ensembl, so I think others will experience this reporting snafu.

[bontus]

"others will experience this reporting snafu"
Indeed, I can confirm this issue after switching to a GENCODE based annotation for my newest run. The header structure is as follows (UCSC):

hg38_wgEncodeGencodeCompV24_ENST00000371007.6 range=chr1:67092165-67231852 5'pad=0 3'pad=0 strand=- repeatMasking=none

[MiladAD123NGS]

Please answer this. I have the same problem.