Protocol Online logo
Top : Forum Archives: : Bioinformatics and Biostatistics

Error message when running blastpgp (from NCBI PSI-BLAST) - (Oct/15/2006 )


I'm trying to run the local version of ncbi psi-blast (blastpgp) using a self-generated database and a fasta query file. I get the following error messages repeatedly while the program runs:

[NULL_Caption] ERROR: ncbiapi [000.000] BioseqFindFunc: couldn't uncache
[NULL_Caption] ERROR: ncbiapi [000.000] ObjMgrNextAvailEntityID failed with idx 2048

I'm using the following command line:

blastpgp -d db_file -i query_file -e 1.0 -m 8 -o output_file -I T -a 2 -j 4 -h 0.001

While experimenting I've noticed that when replacing the option "-m 8" (tabular output format) with "-m 0" (default output format), the errors go away.

Does any one has any information about what could be causing these errors?
Thanks in advance.



Are you sure your input file is in FASTA format? There might also be formatting issues with the database indices (multiple entries with identical tag lines??).

Try formatting the database using formatdb and then run psiblast

Hope this helps

Good Luck! smile.gif

I used formatdb for the database.

As for the input files, it is recognized as fasta format by other programs, so I'm guessing its ok.

Since changing the "-m" options eliminates the errors (-m affecting just the format of the output file) I don't think the problem is with the input. I have wild guess that the problem is related to memory issues, since changing other parameters such lowering the e-value threshold reduces the number of times the errors are repeatedly printed.

However, I still have no clue how I can avoid the errors and still get the outpit format I need...


If you suspect it is a memory thing try dropping the number of cycles to 1 and see what happens - especially with your -e set to 1.0... what spec computer are you running this on?