I have just installed the Blast+ tools from the Galaxy tool shed and tried to make a Blast DB from a fasta file in history. The tool runs ok and indicates that a blastdb has been created in: /storage/home/galaxy/production/galaxy-dist/database/files/085/dataset_85192_files/blastdb However there is no directory dataset_85192_files. I have blast installed and I'm able to run makeblastdb on the command line. Do I have to make any other modifications before I can use Blast tools? Thanks for your help Shaun Webb University of Edinburgh -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
On Wed, Dec 5, 2012 at 2:37 PM, Shaun Webb <swebb1@staffmail.ed.ac.uk> wrote:
I have just installed the Blast+ tools from the Galaxy tool shed and tried to make a Blast DB from a fasta file in history. The tool runs ok and indicates that a blastdb has been created in:
/storage/home/galaxy/production/galaxy-dist/database/files/085/dataset_85192_files/blastdb
There should be a collection of files named blastdb.* in the folder /storage/home/galaxy/production/galaxy-dist/database/files/085/dataset_85192_files/
However there is no directory dataset_85192_files. I have blast installed and I'm able to run makeblastdb on the command line.
Do I have to make any other modifications before I can use Blast tools?
I'm unsure what could be wrong. Do the other BLAST+ tools seem to work (e.g. blastp)? Are all the BLAST tools configured the same way in terms of universe_wsgi.ini and their job runners - or do you have some running on the cluster and some on the head node? My guess is something about the mounted storage, but then I would have expected problems with other tools as well. Peter
Thanks for your reply Peter The other tools work fine. I noticed that the makeblastdb process was still running on the server even after Galaxy said the job was complete, but still no files appeared in this directory. I have killed that process for now and will play around with it and see if I can figure it out. I'll let you kno if I find something. Shaun Quoting Peter Cock <p.j.a.cock@googlemail.com> on Wed, 5 Dec 2012 15:44:47 +0000:
On Wed, Dec 5, 2012 at 2:37 PM, Shaun Webb <swebb1@staffmail.ed.ac.uk> wrote:
I have just installed the Blast+ tools from the Galaxy tool shed and tried to make a Blast DB from a fasta file in history. The tool runs ok and indicates that a blastdb has been created in:
/storage/home/galaxy/production/galaxy-dist/database/files/085/dataset_85192_files/blastdb
There should be a collection of files named blastdb.* in the folder /storage/home/galaxy/production/galaxy-dist/database/files/085/dataset_85192_files/
However there is no directory dataset_85192_files. I have blast installed and I'm able to run makeblastdb on the command line.
Do I have to make any other modifications before I can use Blast tools?
I'm unsure what could be wrong. Do the other BLAST+ tools seem to work (e.g. blastp)? Are all the BLAST tools configured the same way in terms of universe_wsgi.ini and their job runners - or do you have some running on the cluster and some on the head node?
My guess is something about the mounted storage, but then I would have expected problems with other tools as well.
Peter
-- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
On Wed, Dec 5, 2012 at 4:27 PM, Shaun Webb <swebb1@staffmail.ed.ac.uk> wrote:
Thanks for your reply Peter
The other tools work fine. I noticed that the makeblastdb process was still running on the server even after Galaxy said the job was complete, but still no files appeared in this directory. I have killed that process for now and will play around with it and see if I can figure it out. I'll let you kno if I find something.
Shaun
OK... this could be a Galaxy issue rather than a tool issue. A few more details might help. For example, we run our Galaxy on a server which (currently) contains the storage on that server (on a DAS), and this is exported to our cluster. I therefore run IO-heavy CPU-light jobs on the Galaxy server itself (e.g. makeblastdb) whereas all the BLAST search jobs are on the cluster. The point about this is heavy-IO over a network storage device introduces more points for failure. In this case perhaps there is a time lag before the Galaxy server can see the new files makeblastdb wrote? Is there anything that might help in the Galaxy paster.log file? Peter
participants (2)
-
Peter Cock -
Shaun Webb