Closed milnus closed 4 years ago
Hi Magnus,
PIRATE runs on either complete or draft genomes. It looks like it isn't finding any CDS features in the GFF files. How were your genomes annotated and what input did you provide to PIRATE?
S
Hi Sion, thank you for the answer. I use GFF3-files provided by Prokka.
The stdout tell me that something(- Loci file contains 17390 loci from 10 genomes.
) is recognised and it match the expected number of loci.
And it can run the initial CD-Hit clustering, which I would guess wasn't possible without any CDS. However, I can see that all the loci are recognised as core, which leaves nothing for the MCL clustering. That might be what is wrong. By adding in a complete genome I add in non-core genes, which makes everything run smoothly.
Hi Magnus,
I haven't encountered that situation before! All of your genomes must be >98% identical. If you provide PIRATE with the option --pan-opt "--cd-low 100" it should change the lowest CD-HIT threshold to 100% identical. If you have any variation in your genome it will be reflected in the outputs. Your should also change the step sizes with -s to reflect some higher BLAST thresholds (add 98,99,100 to the default range, see help -h).
Very odd! You definitely caught my scripts out there :) S
I am trying to construct a pan genome using genomes that consist only of contigs. However, pirate gives me the following error
fail_text.txt:
When I add a complete genome into the pool of genomes, on which the pan genome is constructed, everything works fine. Is this a feature or a bug, and do you have an explanation to why this arises?