Restricting database by taxa in local blast searches

Question

[dimitrischat]

Hello, i am trying to do a blastx search locally but only for plants. How do i restrict it only searching plants? -taxids ? but how do find i all taxids for plants? or for whatever someone needs to do a search ? thanks

[lieven.sterck]

which version of blast are you using? (== can you already make use of the new dbv5 format of blast DBs?)

[dimitrischat]

the version is 2.8.1+ (no idea)

[gb]

You could download taxonomic information here: ftp://ftp.ncbi.nih.gov/pub/taxonomy/new_taxdump/new_taxdump.zip

To get the taxids you can filter the rankedlineage file.

EDIT: Changed the link

Answer 1

[GenoMax]

I got the following answer from NCBI for how to do this with new v5 blast indexes. You may need to upgrade to latest blast+ if v.2.8.1 is not compatible with v5 indexes.

BLAST+ package includes a script that allows you to extract taxid of interest (e.g. for bacteria below, use taxID you are interested in).

$ sh get_species_taxids.sh -t 2 > bacterial.ids

This dumps all the taxids under taxid 2 into the file bacterial.ids. You can then run blast to limit the search to those id's:

$ blastn -db nt_v5 -query test.txt -out out.put -taxidlist bacterial.ids

[dimitrischat]

thanks a lot for your input. Just wanted to ask, because there are these taxid available: Archaea, Bacteria, Eukaryota, Viruses, Other, Unclassified. How can i specify plants ?

[GenoMax]

Viridiplantae (green plants, taxid 33090) is the top of the hierarchy. Narrow down further as needed.

[dimitrischat]

so i do this ?

sh get_species_taxids.sh -t 33090 ?

[GenoMax]

Yes. Embryophyta (taxid: 3193) are land plants (if you want to narrow down further).

[dimitrischat]

thank you for your input! Much appreciated

[dimitrischat]

-db database_name = the folder i downloaded from ncbi database ( which i downloaded from here = update_blastdb.pl --decompress nr [*] ) ?

[GenoMax]

-db should point to the basename (nr) of the database you are going to search against.

[dimitrischat]

i did

-db '../ncbi_blast_database/nr'

and

-taxidlist '.../ncbi-blast-2.10.0+/bin/plant.ids'

hope this works

[dimitrischat]

i got nr00 up to nr38, all files about ~200gb. I need all right? i mean when i use -db ../ncbi_blast_database/nr , it searches all? but with -taxidlist it searches only this taxa within all nr files?

[GenoMax]

Correct. You need all nr files. With -taxidlist blast should report only the hits you are interested in.

[dimitrischat]

using as database the uniprot_sprot.fasta, can i use the -taxidlist plants.ids ( sh get_species_taxids.sh -t 33090 > plants.ids ) ? or it doesnt make any sense? And if i cant, how can i do that with the uniprot_sprot database?

[GenoMax]

Yes that should work. You mean with swissprot DB correct? That is all NCBI has.