This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Confused about which database to use - nt or nr?

Hi all,

I work with metagenomic and metatranscriptomic data. I know the nt/nr database is for nucleotide reads which my data is all in nucleotide form (not in protein form so I wouldn't be using blastp). But when I do a local blast, I am just wondering how to know if I should use nt or nr. Because with blastp, you only use nr for protein nucleotide sequences. If I have metatranscrptomic reads, I did use blastn against the nt database and I got a lot of hits so I felt that was okay.

When I use FragGeneScan to extract protein reads from the RNA sequences I did use nr database and just wondering if this makes sense or should I have used the nr database from the start?

I've read a lot on this on NCBI but it is very clear to me in practise.

blast
• 7,251 views • link
updated by 5heikki 5heikki • written by DNAngel DNAngel

2 answers

Because with blastp, you only use nr for protein nucleotide sequences.

There is no such thing as protein nucleotide sequences. The sequences are either nucleotide (nt) or protein (nr). blastn is for comparing nucleotide queries to nt database, and blastp for comparing proteins to nr database.

• 0 views • link
written by Mensur Dlakic Mensur Dlakic

...or blastx for translated dna against the nr

• 0 views • link
written by Jeremy Leipzig Jeremy Leipzig

... or tblastn for protein against translated DNA in nt. I don't think the original question was intended for all BLAST flavors and all search permutations.

• 0 views • link
written by Mensur Dlakic Mensur Dlakic

"translated DNA:protein (BLASTX) searches are far far more sensitive than DNA:DNA searches" - Bill Pearson

if you're happy with your blastn results that's good but blastx to the nr would be advisable

• 0 views • link
written by Jeremy Leipzig Jeremy Leipzig

If you have transcripts (3 potential reading frames) it's rather wasteful to do blastx with its 6 reading frames. It's far more computationally efficient to predict proteins and then do blastp. No loss in sensitivity. It doesn't really matter if you're querying just a few hundred sequences or whatever but with large sets there will be a huge cost..

• 0 views • link
written by 5heikki 5heikki

Log in to answer this question.

More posts like this