[maker-devel] Improving BUSCO stats

[Kyungyong Seong]

Hi

The BUSCO statistics obtained from my genome seems to be decent with 97.3%
completeness (-m geno). I am having problems generating genome annotation
sets that show comparable BUSCO completeness (-m prot). Currently,
completeness is around 88%, and iterative MAKER annotation is not
significantly increasing this value.

I started with prot2genome and cdn2genome alignments. I then trained
AUGUSTUS with BUSCO gene sets and SNAP with the predicted gene models with
good quality, and ran MAKER without prot/cdn2genome. The third run was with
newly trained AUGUSTUS and SNAP, which only increased the BUSCO
completeness by 2%. I imagined that single copy orthologs would be well
supported by evidence and may be relatively easy to predict as well. I
wasn't quite sure what is happening. Would you have any advice?

Thank you!
Kyungyong
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20210516/37abb7f0/attachment-0001.html>