Dear Peter,

APOLOGIES for the delayed response

Please see below for details......

On Thu, Aug 1, 2013 at 8:51 AM, Peter Cock <p.j.a.c...@googlemail.com> wrote:
> On Thu, Aug 1, 2013 at 2:47 PM, Rodolfo Aramayo <raram...@tamu.edu> wrote:
>> On Thu, Aug 1, 2013 at 8:40 AM, Peter Cock <p.j.a.c...@googlemail.com> wrote:
>>>>>
>>>>> I've seen that kind of error from NCBI BLAST+ before - do you
>>>>> know which of the BLAST tools was this, and what the query
>>>>> was?
>>>>
>>>> The repo was:
>>>> repos/devteam/ncbi_blast_plus/9dabbfd73c8a/ncbi_blast_plus
>>>>
>>>> The query was:
>>>> Output from velvetoptimiser
>>>> /repos/simon-gladman/velvet_optimiser/43c89d82a7d3/velvet_optimiser
>>>
>>> OK, so this was a nucleotide FASTA file from an assembly, right?
>>
>> Yes
>>
>>> What were you searching with BLAST? e.g. BLASTN against NT,
>>> BLASTX against NR, etc.
>>
>> Search was blastn, task megablast of the assembly as query against an
>> in-house made (and installed into galaxy) of a blastn database for
>> Chromosomes 1 to 7 of Neurospora crassa
>> Database was produced with:
>> makeblastdb -in Ncrassa_wt_Chr01.Chr07_v12 -dbtype nucl -title
>> Ncrassa_wt_Chr01.Chr07_v12 -parse_seqids -hash_index
>> and "installed" into the "blastdb.loc" file as required
>
> Can you run this blastn search 'by hand' at the terminal?

Yes, yes and no

> Does that crash too?

No and I do not know

Please let me explain.
I have set up the user galaxy account such that it has access to a
directory called 'bioinfosoft' where all bioinformatics-related
software is installed, including Blast+
Galaxy has 'access' to this directory and there I have compiled Blast+ 2.2.28+
Now, when I installed Blast+ via the toolshed, the toolshed Blast+
package did download and install what I believe it was the 2.2.26+
version of Blast+?
If I recall correctly, I got an error message that went away while the
Blast+ package was being compiled by the toolshed
The errors I got running Blast did all occur when using that
particular version of Blast+ (2.2.26)
I then removed or uninstalled the packaged and tried to re-install
again, but, as I said, it hanged there forever, installing...
To my surprise when I re-run the jobs, they all worked well. This is,
Blast worked like a charm BUT it did because it is NOT using the Blast
packaged that was installed by the toolshed but the Blast package I
compiled myself (2.2.28+) (SEE LOG AT END OF THIS MESSAGE...)

Comparing the directory of galaxy with those of my backups I can see
that the packaged Blast 2.2.26+ that was installed by the toolshed in
the directory:

./bioinfosoft/galaxy_dependencies/blast+/2.2.26+/devteam/ncbi_blast_plus/9dabbfd73c8a/bin/

is no longer there..

The directory:

./bioinfosoft/galaxy_dependencies/blast+/2.2.26+/devteam/ncbi_blast_plus/9dabbfd73c8a/

is present but it is empty

Therefore to answer your questions:

> Can you run this blastn search 'by hand' at the terminal?

Yes, I can but I would be using a different version of Blast+

> Does that crash too?

No, because I would be using a different version of Blast+

>
> The error sounds very similar to this report which turned out
> to be due to a corrupted BLAST database - so also try
> recreating your DB:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=644962

Regarding this last comment..basically a corrupt database is not the
problem in this case as demonstrated by the fact that I can run Blast
on those databases

Having said that, yes I have observed that depending on how one runs
the makeblastdb command and depending on the type of fasta files used,
you can end up with corrupt output. Some identifiers and symbols
present in the fasta ID can interfere with makeblastdb (especially
some produced by the JGI)

In summary, yes Blast+ is working but because is not using the
compiled Blast+ binaries installed by the toolshed package

The problem I have is that I would like to be able to properly
uninstall and re-install the Blast+ toolsheed so that things work as
they were originally supossed to

On the other hand, if you guys upgrade the package, this might be a
non-issue as the toolshed would just simply generate new directories...

Last, I noticed the command used in the compilation is different from
the one I normally use

The command I use is:

tar -xzvf ncbi-blast-2.2.28+-src.tar.gz
cd ncbi-blast-2.2.28+-src/c++
./configure --prefix=/usr/local/galaxy/bioinfosoft/blast-2.2.28+
--with-64 --with-bincopy --with-mt --without-debug --with-static | tee
c.log
make | tee m.log
make check
make install

Many Thanks


--Rodolfo

>
> Peter

LOG:

galaxy.jobs.handler INFO 2013-08-05 12:11:30,841 (2405) Job dispatched

galaxy.tools DEBUG 2013-08-05 12:11:31,018 Building dependency shell
command for dependency 'blastn'

galaxy.tools WARNING 2013-08-05 12:11:31,018 Failed to resolve
dependency on 'blastn', ignoring

galaxy.tools DEBUG 2013-08-05 12:11:31,018 Building dependency shell
command for dependency 'blast+'

galaxy.tools WARNING 2013-08-05 12:11:31,018 Failed to resolve
dependency on 'blast+', ignoring

galaxy.jobs.runners.local DEBUG 2013-08-05 12:11:31,099 (2405)
executing: blastn -version &>
/usr/local/galaxy/galaxy-dist/database/tmp/GALAXY_VERSION_STRING_2405;
blastn -query 
"/usr/local/galaxy/galaxy-dist/database/files/002/dataset_2897.dat"
  -db 
"/data/db/fungidb/Ncrassa_OR74A/blastplus_db/Ncrassa_wt_Chr01.Chr07.+ChrM.+rDNA_v12"
-task megablast -evalue 0.001 -out
"/usr/local/galaxy/galaxy-dist/database/files/003/dataset_3596.dat"
 -outfmt "6 std sallseqid score nident positive gaps ppos qframe
sframe qseq sseq qlen slen" -num_threads 8

galaxy.jobs DEBUG 2013-08-05 12:11:31,152 (2405) Persisting job
destination (destination id: local)

galaxy.jobs.runners.local DEBUG 2013-08-05 12:11:32,230 execution
finished: blastn -version &>
/usr/local/galaxy/galaxy-dist/database/tmp/GALAXY_VERSION_STRING_2405;
blastn -query 
"/usr/local/galaxy/galaxy-dist/database/files/002/dataset_2897.dat"
  -db 
"/data/db/fungidb/Ncrassa_OR74A/blastplus_db/Ncrassa_wt_Chr01.Chr07.+ChrM.+rDNA_v12"
-task megablast -evalue 0.001 -out
"/usr/local/galaxy/galaxy-dist/database/files/003/dataset_3596.dat"
 -outfmt "6 std sallseqid score nident positive gaps ppos qframe
sframe qseq sseq qlen slen" -num_threads 8

galaxy.jobs.runners DEBUG 2013-08-05 12:11:32,324 executing external
set_meta script for job 2405:
/usr/local/galaxy/galaxy-dist/set_metadata.sh ./database/files
/usr/local/galaxy/galaxy-dist/database/job_working_directory/002/2405
. /usr/local/galaxy/galaxy-dist/universe_wsgi.ini
/usr/local/galaxy/galaxy-dist/database/tmp/tmp7F7qqE
/usr/local/galaxy/galaxy-dist/database/job_working_directory/002/2405/galaxy.json
/usr/local/galaxy/galaxy-dist/database/job_working_directory/002/2405/metadata_in_HistoryDatasetAssociation_4245_qhcDd5,/usr/local/galaxy/galaxy-dist/database/job_working_directory/002/2405/metadata_kwds_HistoryDatasetAssociation_4245_pKXnUR,/usr/local/galaxy/galaxy-dist/database/job_working_directory/002/2405/metadata_out_HistoryDatasetAssociation_4245_jSXP0f,/usr/local/galaxy/galaxy-dist/database/job_working_directory/002/2405/metadata_results_HistoryDatasetAssociation_4245_WzHo1J,,/usr/local/

galaxy/galaxy-dist/database/job_working_directory/002/2405/metadata_override_HistoryDatasetAssociation_4245_OBr8F0
galaxy.jobs.runners DEBUG 2013-08-05 12:11:33,882 execution of
external set_meta for job 2405 finished

___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:
  http://lists.bx.psu.edu/

To search Galaxy mailing lists use the unified search at:
  http://galaxyproject.org/search/mailinglists/

Reply via email to