Problems In Creating Pssm Using Psiblast

Question

[Leo]

Hi, Everyone.

Recently I am using PSIBlast to create PSSM.

Command seems like this:

psiblast -subject oneseq.fasta -inmsa alignment.fasta -outascii_pssm pssm.txt

The problems is

Error: NCBI C++ Exception: "/am/ncbiapdata/release/blast/src/2.2.25/IntelMAC-universal/c++/GCC401-ReleaseMT--IntelMAC-universal/../src/util/sequtil/sequtilconvertimp.cpp", line 225: Error: ncbi::CSeqConvert_imp::Convert() - Unknown conversion.

I have tried several times and did not figure out this problem.

Could anyone can help me to solve this problem?

Great thanks.

[Michael Schubert]

Can you copy your oneseq.fasta and alignment.fasta into pastebin?

Answer 1

[Hamish]

The error appears to be related to handling the input sequence or alignment. From the file names being used my guess is that the alignment is not in the required format, see "4.8.1 Multiple sequence alignment" for details of the required format.

Given this is a "Convert" error check the sequences for non-sequence characters, I suggest using the '-v' option to 'cat' to check for any non-printable characters in the data:

cat -v oneseq.fasta
cat -v alignment.fasta

To check that NCBI BLAST+ is working correctly try using the example sequence and alignment provided in "3.3 Jump Start PSI-BLAST Search Using a Multiple Sequence Alignment", from the "Standalone PSI/PHI-BLAST: blastpgp" documentation for classic BLAST. The following works as expected for me using NCBI BLAST+ 2.2.25 on a Linux system:

psiblast -subject seq1 -in_msa align1 -out_ascii_pssm pssm.txt

If this doesn't work for you then I would suspect there is a problem with the BLAST+ installation or a problem in the MacOS X version of BLAST+ being used. In either case the BLAST folks at NCBI (see BLAST help page) may be more able to assist with the problem.