datasources.txt Uniprot identifiers causes IMS to fail

[bridgedb]

datasources.txt Uniprot identifiers causes IMS to fail

[ariutta]

The error could be located in one of several places. I'll describe what I'm trying to do, and maybe you can help pinpoint the source of the problem.

Goal: convert WikiPathways Xrefs to Miriam/identifiers.org IRIs.

Example: pathway WP3998 has this element in its GPML:

<Xref Database="Uniprot-SwissProt" ID="P24385" />

I thought I could use datasources.txt to map from GPML Database to Miriam/identifiers.org namespace. According to datasources_headers.txt, Database corresponds to datasource_name, which is the first column of datasources.txt. And datasources_headers.txt also says the ninth column of datasources.txt, uri, is the:

Official URI for datasource (e.g., from miriam); or repeat of system_code if unknown or unregistered

These are the potential fixes I see:

1. WikiPathways should only use Uniprot-TrEMBL, not UniProtKB/Swiss-Prot
2. datasources.txt should list urn:miriam:uniprot for UniProtKB/Swiss-Prot
3. identifiers.org should create a new namespace for UniProtKB/Swiss-Prot

Supporting 1), it seems that at WikiPathways, all Xrefs with Database values of UniProtKB/Swiss-Prot have ID values that match the datasources.txt regex for Uniprot-TrEMBL, not the one for UniProtKB/Swiss-Prot.

Supporting 2), identifiers.org states that UniProtKB/Swiss-Prot is an alternative name for namespace uniprot. That means the Miriam/identifiers.org URI for Uniprot-SwissProt is registered. If that's true, then according to datasources_headers.txt, the ninth column should be urn:miriam:uniprot for Uniprot-SwissProt.

Supporting 3), the URI http://identifiers.org/uniprot/CALM_HUMAN fails to resolve. CALM_HUMAN is the example identifier listed for UniProtKB/Swiss-Prot in datasources.txt.

[AlexanderPico]

I think we should reverse the problematic change right away to get IMS working again.

@ariutta Could you use the UniProt-TrEMBL row in datasources.txt for both swissprot and uniprot mappings in the meantime?

Longer term, we could explore other solutions involving miriam updates or adding another column to datasources or converting all swissport refs to uniprot (when using accessions), etc.

[ianwdunlop]

Thanks for the explanation @ariutta. I think the current head in master has the change that I needed for IMS but I wanted to understand a bit more before you released the latest version.
I probably favour option 3 ie asking identifiers.org to add a new UniprotKB/Swiss-Prot namespace.

[egonw]

This is the remaining patch in the OPS master left in the merge and pertains to this issue:

openphacts@fa32572

[egonw]

Closed by 192a18f