NCBI blastp fails with "Invalid choice selection: NCBI-Seqalign::Score.value.real"

Question

[claudius.mueller]

I'm trying to blast using the command line blast+ package from the NCBI. Blasting for a short peptide using the task option "blastp-short":

blastp -task blastp-short -out test.html -query test.fasta -db nr -remote

gives me the following error:

Error: NCBI C++ Exception:
    T0 "/home/clausimu/Downloads/ncbi-blast-2.9.0+-src/c++/src/objects/seq/../seqalign/Score_.cpp", line 90: Error: CScore::C_Value::GetInt(): Invalid choice selection: NCBI-Seqalign::Score.value.real

Anybody have an idea what I'm doing wrong?

EDIT: I believe this might be related to the specific matrix that is used. blastp-short uses PAM30 by default, instead of BLOSUM62. If I blast using blastp without the "blastp-short" task, but using the "-matrix PAM30" option, I get the same error. If I don't use this option, blast runs fine (but won't find any hits, because this matrix is critical for short peptide searches).

Thanks!

[Torst]

Score.value.real suggests maybe you should add an -evalue 1E-3 or something?

As an aside, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3848038/ is a good article on matrices for short queries.

[mrmrwinter]

Could you direct me towards a solution for this error please? I'm receiving the same trying to do a blastn search ----- "blastn -db nt -query testseq.fasta -out ../blasted/testseq.out -remote"

I dont know enough about blast or matrices to interpret your edit, sorry

Answer 1

[joines]

I suspect you have encountered the same bug that I ran into. I opened a ticket with the NCBI helpdesk and received this response:

"Date: Thu, 10 Oct 2019 07:40:48 ... Thank you for the report. There is a bug in 2.9.0 when using the '-remote' option. It should work in 2.8.1, and it has been fixed for the upcoming 2.10.0 release, which should be out in 2 or 3 weeks."