Hi,
I have RNA-Seq transcripts peptide sequences, I performed blast annotation against NR and Swissprot database using standalone blast in linux. I got nr and Swissprot databases from NCBI.
Now I need to perform blast annotation against Gene Ontology (GO), KEGG, and EuKaryotic Orthologous Group (KOG) databases. Where can I get fasta files to build blast databases for Gene Ontology (GO), KEGG, and EuKaryotic Orthologous Group (KOG) in order to annotate genes against aforementioned databases using standalone blast.
Thanks a lot!
•
written
by
Bushra Bibi
Don't know about GO and KEGG, but pretty sure there is no FASTA database for KOG.
BLAST annotation is decidedly old-school, which I don't mean as a negative because I do many old-school things. However, it is very inefficient and often inaccurate to annotate sequences on a large scale using BLAST. If you insist on using a BLAST-like approach, I suggest rpsblast instead and use the CDD database:
A proper way to do this would be by using a combination of hidden Markov models (HMMs) databases. Rather than getting thousands of BLAST hits, some of which might be uninformative, HMMs will give a smaller number of hits that are clearly annotated. What I suggested above with rpsblast is an approximation of a proper approach, which involves getting HMMer tools installed and using them against HMM databases:
Thank you so much! I will use rpsblast. And what do you you suggest about diamond blast? Also which approach do you think should be used instead of blast annotation.
DIAMOND is faster than BLAST and a little less sensitive, so all the issues I outlined for BLAST apply.
I already gave you two different approaches to use instead of BLAST annotations, and even described their pros and cons. Searches against HMM databases will be more sensitive, but rpsblast may be easier to implement since you must already have a program if you were running BLASTp searches.
Log in to answer this question.
