Add stopgap fix for Stockholm format alignment export

[twalsh-ebi]

Description

Because of a bug in the Stockholm alignment writer in BioPerl 1.6.x, homologue alignment downloads in Stockholm format have sequence identifiers that run on into the accession field:

# STOCKHOLM 1.0
#=GF ID NoName
#=GS ENSMUSP00000089009_Mmus/1-3623AC unknown
#=GS ENSMUSP00000087102_Mmus/1-435AC unknown
#=GS ENSMUSP00000152236_Mmus/1-540AC unknown
#=GS ENSMUSP00000099398_Mmus/1-1385AC unknown

This prevents the alignment from being loaded into (for example) Biopython, making it more difficult to process downloaded Stockholm files automatically.

This bug was fixed in bioperl-live PR 122, so code incorporating a bugfix is available in the event of an upgrade to BioPerl 1.7.0+.

In the meantime, this PR corrects the issue by fixing the generated Stockholm alignment string. As an illustration:

# STOCKHOLM 1.0
#=GF ID NoName
#=GS ENSMUSP00000089009_Mmus/1-3623 AC unknown
#=GS ENSMUSP00000087102_Mmus/1-435 AC unknown
#=GS ENSMUSP00000152236_Mmus/1-540 AC unknown
#=GS ENSMUSP00000099398_Mmus/1-1385 AC unknown

This makes it easier for Compara team members to test — and Ensembl users to work with — downloaded Stockholm files.

Views affected

This affects export of Stockholm format alignments of homologues.

To see the effect of this PR, try previewing or downloading Stockholm alignments for the following two examples:

• Human BRCA2 orthologues: sandbox vs live site
• Mouse Cntnap1 paralogues: sandbox vs live site

Possible complications

As these changes only have an effect on the GS lines of Stockholm-format alignments in which the sequence identifier runs on into the accession field, no complications are expected.

Merge conflicts

None detected.

Related JIRA Issues (EBI developers only)

N/A

[azangru]

@jyothishnt is wondering if this is always "AC unknown"? In the PR to biomart-perl, AC is hard-coded, but "unknown" is not.

[twalsh-ebi]

Hi @jyothishnt and @azangru.. With the current Ensembl codebase, I believe this suffix is indeed always "AC unknown" in practice.

The $acc is set from the $seq->accession_number in Bio::AlignIO::stockholm, and a sequence accession_number is not set anywhere in the Compara codebase, though it is accessed here.