BLASTP problems

[heculase]

Hi all,

I blasted a fasta file with 15000 sequences on a linux machine using the following command:

blastp -query protein.fasta -db nr -evalue 1e-6 -num_threads 64 -outfmt 5 -max_target_seqs 5 -out output.xml

However I could only obtain results for the first 405 sequences. The program run off at some timepoint (about 2 hrs on our cluster). What would be the problem?

Thanks a lot!

[GenoMax]

It sounds like a standalone machine based on your first sentence but then you refer to a "cluster" later on. Are you running into some sort of limits set by your system administrators on system resource usage (or something as simple as a disk partition filling up)? Are there any log files (either from blast or queuing system from your cluster)?

If you have 15000 sequences to blast then you may want to split those into several smaller jobs (if you are running on a cluster).

[heculase]

It sounds like a standalone machine based on your first sentence but then you refer to a "cluster" later on. Are you running into some sort of limits set by your system administrators on system resource usage (or something as simple as a disk partition filling up)? Are there any log files (either from blast or queuing system from your cluster)?

If you have 15000 sequences to blast then you may want to split those into several smaller jobs (if you are running on a cluster).

Thank you very much GenoMax.

It is a standalone machine, not a cluster -- mistaken.

The system administrator said neither disk space nor CPU usage was limited, so I was trying to split the file into minifiles containing 500 sequences. Blasting one of them took around 2.5 hrs and had ~400 sequences aligned. So I guess it is like "sooner or latter the program will exist for itself for unknown reason"?

I am now splitting the fasta to 200 sequences per file and will try it.

[GenoMax]

Are you getting the result in a truncated file or is the program throwing an error when it exits?

When something like this happens one thing to check is if the program is writing intermediate files to /tmp and to see if that fills up over time. You may need to ask the admins to verify that.