Closed ohickl closed 2 months ago
Hi @ohickl , thank you very much for reporting! I just took a deeper look into this and found a wrong assumption that I made for the CRISPR parser of PILER-CR.
I wrongly assumed that all CRISPR spacers are of length N>=10
. But, in your example, the CRISPR spacer is of length 8. Because of this, the regex didn't match to the result file. I pushed a PR with a fix. I'll run some tests and release a patch release, soon - hopefully by the end of this or next week.
Thanks again!
Hi Oliver, I performed quite a few runs now on metagenomes of various sizes with the modified Bakta version from #289 and had barely any problems except for one sample and specifically with one contig. There, a PILER-CR predicted spacer seq doesnt match the seq in the fasta at the position parsed by Bakta.
Here the stdout from a run with regular Bakta and just the one contig:
I do find the seq in the fasta at index -6 from
crispr_spacer['start']
, but am not sure how/if its relevant:Im not sure, if the problem is PILER itself or the result parsing, but I get this with all Bakta versions.
I attached the fasta of the contig and a manual PILER run with the same params used in Bakta.
Let me know, if I should get in touch with the PILER people directly for this.
Best
Oskar piler_test_out.txt piler_test_contig.txt