Blast p error with database definition

Question: Blast p error with database definition

3.0 years ago by

olivier.claude • 30

France

olivier.claude • 30 wrote:

Hello,

I am trying to run a blast with my datas.

I installed Blast+ package from the tool shed.

I downloaded two different files from the ncbi:

-swissprot (from the FASTA section) -> called swissprot-15-11-25

-swissprot with the different files ( .phr, .pin . pnd ... ) -> called swiprot.00

I have defined in the blastdb_p.loc as follow:

swissprot-15-11-25 <tab> swissprot-15-11-25 <tab> /home/olivier/galaxy/blastDB/swissprot-15-11-25
swissprot.00 <tab> swissprot-ncbi.00 <tab> /home/olivier/galaxy/blastDB/swissprot.00

I Tried to run the blast with the etwo different databases and it gave me the error code :

Fatal error: Exit code 2 ()
BLAST Database error: No alias or index file found for protein database [/home/olivier/galaxy/blastDB/swissprot.00] in search path [/home/olivier/galaxy/database/job_working_directory/000/380::]

I runned the command line :

formatdb -i swissprot-ncbi-15-11-25 -p F -n "swissprot-ncbi-15-11-25" -v 2000

but it didn't changed anything.

Who did i missed?

thank you.

Olivier

error blast+ database • 860 views

ADD COMMENT • link •

modified 3.0 years ago • written 3.0 years ago by olivier.claude • 30

3.0 years ago by

Jennifer Hillman Jackson ♦ 25k

United States

Jennifer Hillman Jackson ♦ 25k wrote:

/Hello,

The indexes should be downloaded from here: ftp://ftp.ncbi.nlm.nih.gov/blast/db/. The fasta file is not needed and should not be referenced in the loc file. However, you might want a copy that matches the index file anyway. Once downloaded, make sure these are fully uncompressed. Then make sure that the last column in the file includes a path that ends with a basename that is an exact match for the first part of the index names. Your loc seems to be missing a line in that format (although it is a bit difficult to tell from the other entries exactly which directory contains the index files). You only need one line per database.

For example, if the indexes are in here and the basename is swissprot:

/home/olivier/galaxy/blastDB/swissprot-15-11-25/swissprot.[number].[index]

The correct loc entry path would be:

/home/olivier/galaxy/blastDB/swissprot-15-11-25/swissprot

You can use whatever you want for the first two columns - these are labels. The first must be one "word" (no spaces), the second can include multiple words separated by spaces (but no internal tabs).

See the top of the loc file example for the full details. This covers the same information but in more detail. After making changes, restart the server.

If for some reason you do want to create your own, note that the formatdb line-command is incorrect for protein indexes. Specifically, "-p F" indicates that the input file is in nucleotide format, not protein.

Most likely, the loc file path change is enough. Adjust my example to the actual path to the indexes in your directory structure.

Thanks, Jen, Galaxy team

ADD COMMENT • link written 3.0 years ago by Jennifer Hillman Jackson ♦ 25k

3.0 years ago by

olivier.claude • 30

France

olivier.claude • 30 wrote:

it works !

thanks

ADD COMMENT • link written 3.0 years ago by olivier.claude • 30

Super, very glad you have this going! Jen

ADD REPLY • link written 3.0 years ago by Jennifer Hillman Jackson ♦ 25k

Please log in to add an answer.

Similar posts • Search »