Problems with blastx and thousands of equences

[angeloulivieri]

Hi all, I'm running a blastx in my server. Iw had works everytime but now for a large file of sequences is doing something strange.

The query is the following:

nohup ../ncbi-blast-2.2.27+/bin/blastx -db ../bin/data/uniprot_kb_2012_06.fasta -query 42000seq.fa -evalue 0.05 -max_target_seqs 5 -outfmt 5 -num_threads 10 -out ouBlastxXML &

where uniprot_kb_2012... is a dataset containing all the protein taken from ncbi.
42000seq.fa is a file containing 42thousands sequences in fasta format

the ouput I want is in XML...

I runned it one week ago and obtained a completely empty file!

Now with nohup it's writing this:

Selenocysteine (U) at position 73 replaced by X
Selenocysteine (U) at position 40 replaced by X
Selenocysteine (U) at position 52 replaced by X
Selenocysteine (U) at position 48 replaced by X
Selenocysteine (U) at position 37 replaced by X
Selenocysteine (U) at position 40 replaced by X
Selenocysteine (U) at position 40 replaced by X

...and other similar lines...

and the xml file is still empty...

What's happening?

The same command on a query of ten sequences works well.

Someone knows where can I been wrong?

bye and thanks
Angelo

[westerman]

Don't know what is wrong but you best bet is to run Blast with a small group of sequences and then put all of the XML files together.

[maubp]

It seems BLAST+ is a bit silly with the XML output and doesn't write it out incrementally.

I personally split the input into batches of 1000 queries (works well for spreading the work over a cluster).

[angeloulivieri]

can the reason be the fact that I use an XML file to extract informations?

[maubp]

can the reason be the fact that I use an XML file to extract informations?

If you use the text or tabular output, you should see results written to the file while BLAST+ is running.

[angeloulivieri]

so only with these options? The 6,7 and 8...
With other types my output will be full only at the end of the computation?

[maubp]

I've only noticed a problem of delayed output with the XML output format.

[angeloulivieri]

Ok. The first time I used blastx with these 40thousands sequences it doesn't give me nothing and for me was very strange. Now I'm trying using a not-XML output file cause I need only to use some bioPerl functions to watch for results.

Thanks

[dariober]

Selenocysteine (U) at position 73 replaced by X
Selenocysteine (U) at position 40 replaced by X
Selenocysteine (U) at position 52 replaced by X
Selenocysteine (U) at position 48 replaced by X
Selenocysteine (U) at position 37 replaced by X
Selenocysteine (U) at position 40 replaced by X
Selenocysteine (U) at position 40 replaced by X

Hi-
Just for information, from http://www.ncbi.nlm.nih.gov/BLAST/blastcgihelp.shtml this is due to

For protein code, U is replaced by X first before the search since it is not specified in any scoring matrices.

so this is nothing wrong. I don't know about the rest...

Best
Dario

[angeloulivieri]

Thank you Dario