blastp multiple local databases error

Question

[dp]

Hi I've come across an error/bug using blastp with multiple local databases. I am blasting against a local copy of nr and an in house database. When I blast against each separately there is no problem, but when I use the option:

-db '/path/to/mydb  /path/to/nr'

I get the following error: BLAST Database error: Error pre-fetching sequence data

Any ideas how to deal with this error?

Thanks!

[GenoMax]

You will need to aggregate multiple databases with blastdb_aliastool as described here.

[dp]

Thanks! I will try that now. Do you know why the -db option doesn't work? This section of the user manual says: "BLAST has been able to search multiple databases since 1997. The databases can be listed after the “-db” argument or in an alias file"

[GenoMax]

Good question.

What blast+ version are you using? Latest version of blast uses v5 format blast indexes. Perhaps you may have to re-index your own database to be v5 format in case you are using fresh nr from NCBI (which is now is v5 format).

[dp]

Thanks! Everything should be v5 - I am using BLAST+ 2.10 and fresh nr. I get the same error using an alias file as you suggested

[GenoMax]

That error seems to come from NCBI's c++ toolkit. Did you compile your own blast? Do you have the latest c++ toolkit?

[dp]

No - I just downloaded the blast+ v2.10 executables for linux

[GenoMax]

You may want to send a helpdesk ticket into blast help on this. Could be a bug that you have uncovered.

[dp]

OK good idea. Thanks!

Answer 1

[dp]

I learned that all dbs must have the same value of use_seqids. Since nr was created using use_seqids, this worked once I re-created my db with it as well.