Preserve original contig names when using Polypolish

[jmtsuji]

Thanks so much for your work on Polypolish, and congrats on its recent publication!

I have a very minor feature request. When I run polypolish, I notice that it always appends _polypolish on the ends of contig names in the FastA file. E.g.,

>contig_1

Becomes

>contig_1_polypolish

Adding the _polypolish suffix to sequence names is nice because it helps the user to track data provenance. However, it can also be a bit cumbersome sometimes, e.g., if the user wants to programmatically compare assembly statistics (pre-polishing) with other analyses done after polishing. Would it be possible to add a flag to polypolish to preserve the original contig names in the FastA file during analysis?

Thanks!

[jmtsuji]

For those interested, here is a simple workaround to remove the _polypolish suffix on FastA header names using awk:

./polypolish assembly.fasta R1.sam R2.sam 2>polypolish.log | \
  awk '{ if ($0 ~ /^>/) { gsub("_polypolish$", ""); print } \
      else { print } }' \
  > polypolish.fasta

This issue can potentially be closed now given that an awk one-liner can accomplish the desired task.

[rrwick]

Thanks! I've added a new bit on the Polypolish FAQ addressing this case:
Can I prevent Polypolish from changing contig names?

I'll close the issue now 😄

[jmtsuji]

Thanks!

[d-yarmosh]

./polypolish assembly.fasta R1.sam R2.sam | sed 's/_polypolish//g' > polypolish.fasta
If anyone wants a slightly easier to read method to do the same thing as the above awk

Further, it appears that polypolish saves the original headers, provided they aren't separated by spaces -- which Unicycler does and the data after spaces is pretty useful, so you could also do

sed -i 's/ /_/g' assembly.fasta to replace all spaces with underscores and then apply the above sed with a small addition like:
./polypolish assembly.fasta R1.sam R2.sam | sed 's/_polypolish//g' | sed 's/_/ /g' > polypolish.fasta

[hkaspersen]

@rrwick Just to add here, do you have any plans on looking into the above issue that happens when spaces are used in contig names? I was wondering if this was an easy fix within polypolish itself, or perhaps in unicycler!

[rrwick]

Polypolish doesn't current save any of the sequence description strings (anything after the first whitespace in the FASTA header), but this could potentially be added. I'll reopen the issue and tag it as an enhancement request. Thanks!

[rrwick]

I have just released Polypolish v0.6.0, which addresses this issue. Contig descriptions are now kept, and polypolish is added to the end of the description instead of to the contig name. See this FAQ entry for an example of how it now works.