ID mapping services

[mihai-sysbio]

There has been a lot work done regarding ID mapping. I'm wondering if BridgeDB has been considered so far, given that is seems to cover a wide range of identifiers and that it has an API that I can see useful for automated workflows:

DrugBank
miRBase Sequence
ChEBI
KEGG Reaction
Illumina
Affy
Chemspider
EcoCyc
RefSeq
Unknown data source
KEGG Drug
UCSC Genome Browser
MACiE
GeneOntology
PDB
Rfam
MetaCyc
Uniprot-TrEMBL
Agilent
Guide to Pharmacology
PubChem-compound
Enzyme Nomenclature
HMDB
Reactome
Rhea
Ensembl
OMIM
LIPID MAPS
EPA CompTox
WikiGenes
InChIKey
ChEMBL compound
KNApSAcK
Wikidata
Entrez Gene
HGNC
UniGene
KEGG Compound
Complex Portal
CAS

I took the freedom to cross out those which did not sound useful and highlight the ids already used - please consider this list a work-in-progress and edit this message as you see fit. It's sad to see that MetaNetX is not on this list at the moment.

[haowang-bioinfo]

has anyone tested the performance of this API, if it's updated with the various data sources regularly?

[haowang-bioinfo]

By checking BridgeDB webservice, the following ids can be mapped to Rhea:

• Enzyme Nomenclature (E)
• EcoCyc (Eco)
• MACiE (Ma)
• KEGG Reaction (Rk)
• MetaCyc (Mc)
• Reactome (Re)

[mihai-sysbio]

The remaining question is - what are the benefits of using this in addition to MNXref?

[haowang-bioinfo]

what are the benefits of using this

Very good question that had been explained mostly in #224

Here I'd like reiterate some key points:

• Rhea is a curated and reliable data source
• It is based on the ChEBI compound database that has been tightly integrated with HumanGEM and Met Atlas
• Rhea has close associations with UniProt, with which functional annotation is manually curated and appears to be particularly useful in the recent GPT-assisted curation
• The mapping of Rhea ids will drive forward HumanGEM improvement as the evolving of Rhea

[haowang-bioinfo]

Besides BridgeDB, an alternative implementation (#506) is to map to Rhea ids directly via rhea-db.org/mapping

[haowang-bioinfo]

it might be good, and possible for now, to consistently map from Ensembl to Uniprot/Siwssport ID in the genes.tsv at some point.