Entering edit mode
5.7 years ago
s102060125
▴
10
Hi biostar users,
I am using blast2go to do functional annotation on my newly assembled genome. I blasted my predicted genes against the latest uniprot database on a local server. After importing the blast results into blast2go and mapped them to GOs, I found that the mapping rate was pretty low. I got about 8.6k genes with blast result, but only 1.6k were actually mapped. Is this a usual mapping rate, or I did something wrong? Thanks a lot.
Since you did get some results back in general you must have done the process right. As the post stands there is not enough information in your post to give you any useful pointers.
It would be helpful to add what kind of genome this is, how did you assemble it and predict genes? Have you considered blasting against
nror against a related genome to see if that works better.Thanks for your reply. My genome is a de novo assembled insect genome. The gene sets were generated by the EVM pipeline and manually curated. BUSCO values showed that the gene set is quite complete compare to published relative species. I also annotate the gene set with interproscan implemented in blast2go. Around 95% of the seqs have interproscan hits and ~70% can be assigned to GO. That's why I found it weird to have a blast mapping rate so low. I will try the nr db, but I doubt that will make much difference.
If your insect species is specific/under represented in uniprot it would not be unusual to see low mapping (in human curated part). Did you use both
swiss-protandtrembl?If you have a close relative in the genome databases then perhaps using the proteome from that species would make most sense (instead of
nr).