Results form local blast installation

[JdeBruin]

Hi,

I have a local BLAST installation on our cluster, I aslo have the nr database the uniprot and pfam. Currently I have a dataset that when I blast against the nr or uniprot tremble I get no results returned, but when I use the online blast tool I get results, any ideas why this would be happening?

Thanks!

[WhatsOEver]

Is the latest blast version installed?
Is the downloaded database up to date?
Did you use identical parameters? Can you post your command and described which parameters you changed using the online tool?

[JdeBruin]

This is my blast command:
~/apps/blast/bin/blastn -num_threads 15 -query query.fa -db /scratch/sysusers/jurgens/DB/nr -out qeury.blast.nr.out -outfmt "6 qseqid sseqid evalue qstart qend sstart send qseq sseq length mismatch"

Version :
blastn: 2.2.29+
Package: blast 2.2.29, build Mar 18 2014 08:48:30

I did not change anything on BLAST website just run BLASTN to see if I get any results.

[WhatsOEver]

Ok, just to make it clear (I have seen a couple of people making this mistake):

When you say you just run blastn, you mean you click blastn, paste the sequence and go or do you paste the sequence and switch to "Optimize for Somewhat similar sequences (blastn)" under the "Program Selection" point.
Because the standard setting for the webtool is megablast which gives often different results.

What about your local database - is it up to date? I wouldn't have any other explanation for the differences as the algorithm and parameters are identical for local and online blastn.

[GenoMax]

@JdeBruin: I assume you are referring to BLAST @NCBI. As WhatsOEver has already said, click on the "Algorithm parameters" at NCBI and then compare the default settings (taking into account choice of blast method you select) to the command line ones you are using. They may not be the same.

As for the local command line blast have you been able to get results from other searches?

[JdeBruin]

Yes I am referring to BLAST@NCBI. I have selected for blastn as this is what I have been using on the command line. I have had different dataset produce results just this particular set does not. I will update my nr database as wells as the blast tool set.

[kmcarr]

You may have a database type mismatch. The NCBI "nr" database is a protein (amino acid sequence) database; you can not run blastn with a protein database. I'm surprised your command didn't throw an error if your were trying to run blastn against nr downloaded from NCBI. The database you want to use for a local blastn run is "nt".

[WhatsOEver]

You may have a database type mismatch. The NCBI "nr" database is a protein (amino acid sequence) database; you can not run blastn with a protein database. I'm surprised your command didn't throw an error if your were trying to run blastn against nr downloaded from NCBI. The database you want to use for a local blastn run is "nt".

Yes, that's a good point. When blasting a nucleotide against protein database you will not necessarily get an error as the program just tests for not allowed characters and A,C,G,T are also valid amino acids. The other way around, you will get an error.
The description on the blast website is, however, also a little misleading, as "nucleotide collection (nr/nt)" is the standard db for blastn. This just refers to its partial non-redundance.