database_formatter fails when non-standard but NCBI-like IDs are used

[wwood]

When formating a database where there is already a ncbi-like with a vertical bar like below, but is not recognized by NCBI's makeblastdb, creating the databases fails:

>psu|MAL13P1.200 | organism=Plasmodium_falciparum_3D7 | product=mitochondrial ribosomal protein L49 precursor, putative | location=Pf3D7_13:1587223-1588069(-) | length=111

Excerpt:

$ bin/database_formatter
...
I, [2011-12-16T14:07:39.806625 #15613]  INFO -- : Will run: /home/ben/bioinfo/ncbi-blast-2.2.25+/bin/makeblastdb -in /home/ben/phd/seqserv/db/PfalciparumAnnotatedProteins_PlasmoDB-8.1.fasta -dbtype prot -title "PfalciparumAnnotatedProteins_PlasmoDB-8.1.fasta" -parse_seqids


Building a new DB, current time: 12/16/2011 14:07:39
New DB name:   /home/ben/phd/seqserv/db/PfalciparumAnnotatedProteins_PlasmoDB-8.1.fasta
New DB title:  PfalciparumAnnotatedProteins_PlasmoDB-8.1.fasta
Sequence type: Protein
Keep Linkouts: T
Keep MBits: T
Maximum file size: 1073741824B

No volumes were created because no sequences were found.

Error: NCBI C++ Exception:
    "/am/ncbiapdata/release/blast/src/2.2.25/Linux32-Suse-icc/c++/ICC1010-ReleaseMT--Linux32-Suse-icc/../src/objects/seq/../seqloc/Seq_id.cpp", line 1637: Error: ncbi::objects::CSeq_id::x_Init() - Unsupported ID type psu
...

One workaround might be to format the databases by hand not using the -parse_seqids flag, another is to change the underlying fasta file.

[yannickwurm]

I don't understand. What is special about the database?

[wwood]

Sorry - I pressed submit when I meant to press preview creating the issue - updated. I don't think this is a big priority.

[yannickwurm]

shit. OK well I think this is way beyond our mandate.

We could catch the exception and tell the user to check the IDs. And similarly add a note to the FAQ.

[yeban]

Fixed in BLAST 2.2.30+. Fixed in SequenceServer via b1776df.