I'm using velvet to align given reads of RNA to given CDSs (i.e. coding areas and genes) of an organism, so I can generate gene-expression profiles. But after using velvetg out-dir/ -alignment yes, velvet produces contig-alignment.psa which is a file in a strange format. The file for every contig contains zero or more records of 7 integers, and I cannot understand their meaning.

The following is an excerpt of the file, and you can see the entire file here:

1    193    1939    1939    6   194 2
195 244 1   1   6   22  71
1   84  1   1   6   59  142
86  170 1935    1935    6   22  106
172 285 1935    1935    6   108 221
1   98  2   2   6   1   98
100 321 2   2   6   100 321
334 415 1204    1204    6   1   82
1   23  3   3   6   84  106
  • $\begingroup$ Is that the only file produced? The .psa format is not like what you show (I think, anyway, I haven't used it before but that's what I could find). Are there really no other output files? If not, it sounds as though that might be a bug. $\endgroup$
    – terdon
    Jul 8 '17 at 15:21
  • $\begingroup$ Of course there are other files that also appear while using velvet without the option -alignment yes. But related to contig alignment I can only see this file $\endgroup$
    – hhoomn
    Jul 8 '17 at 15:40
  • 3
    $\begingroup$ I don't think velvet is doing what you think it is. Velvet is a genome assembler. To quote the velvetg command help: "-alignments <yes|no> : export a summary of contig alignment to the reference sequences (default: no)" If you want to align RNA-seq reads to a transcriptome and get "gene-expression profiles" use a tool such as bowtie2 or Salmon. $\endgroup$
    – ithinkiam
    Jul 11 '17 at 11:45
  • $\begingroup$ Thank you, I used Last aligner, but the reason I wanted to use velvet was because It generates coverage ratio that could indirectly imply gene expression. $\endgroup$
    – hhoomn
    Jul 11 '17 at 14:09
  • $\begingroup$ @hhoomn It does in a very roundabout way and it won't give any measure of relative expression between genes. I think you need to explain what exactly it is you want to do in terms of a biological question. It seems to me you're asking the wrong question, here. $\endgroup$
    – ithinkiam
    Jul 13 '17 at 9:05

It looks like you are trying to do a transcript isoform expression analysis. Velvet is a genome assembler, and really not designed for this task. The tools you are looking for are kallisto and salmon.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.