Create infrastructure and SOP for mapping chemicals to CHEBI & swissLipids for generation of chemical measurement terms

[dosumis]

Background - The GWAS catalog needs to curate associations between variants and changes in the levels of many metabolites in, for example, blood. Studies often measure large numbers of metabolites and record these using standard chemical identifiers. We need to map these in to chemical ontologies that we can use - this means CHEBI right now, and possibly SwissLipids in the near future. Both of these ontologies are rich sources of ID mappings.

Proposed strategy - ensure that chemical ID mappings from CHEBI & swissslipids are available in OxO (loaded as SSOM) and that they can be used easily for mapping lists of IDs from GWAS curators to CHEBI/SwissLipids. From there we need SOPs/pipelines to get these IDs into TSVs for generation of OBA terms - with the OBA terms added to lists for import to EFO + a mechanism to inform GWAS curators of term availability.

Potential issue: Chemicals not in CHEBI. Can we devise a strategy for mapping up? Worth discussing with experts on CHEBI channel on OBO foundry.

As well as clear SOPs - all involved in mapping and term generation should have a good understanding of the whole process of generation of and use of mappings so that they can take ownership in future.

[dosumis]

Test: Sheet 2 on this table has a list of metabolites measured in a GWAS study. https://static-content.springer.com/esm/art%3A10.1038%2Fs41588-022-01270-1/MediaObjects/41588_2022_1270_MOESM4_ESM.xlsx

The table is very rich in chemical identifiers:

Metabolites	Compound ID (Metabolon)	CHRO_LIB_ENTRY_ID	COMP_ID	LIB_ID	SUPER_PATHWAY	SUB_PATHWAY	PATHWAY_SORTORDER	TYPE	INCHIKEY	SMILES	CAS	CHEMSPIDER	HMDB	HMDB_curated	KEGG	PUBCHEM	PUBCHEM_curated
S-1-pyrroline-5-carboxylate	X35	166164	42370	400	Amino Acid	Glutamate Metabolism	62	NAMED	DWAKNKKXGALPNW-REWHXWOFAV	OC(C1CCC=N1)=O	2906-39-0	10140206	HMDB0001301	HMDB0001301	C04322	11966181	11966181
spermidine	X50	157851	485	402	Amino Acid	Polyamine Metabolism	545	NAMED	ATHGHQPFGPMSJY-UHFFFAOYAK	NCCCCNCCCN	124-20-9	1071	HMDB0001257	HMDB0001257	C00315	1102	1102
1-methylnicotinamide	X55	155829	27665	400	Cofactors and Vitamins	Nicotinate and Nicotinamide Metabolism	4316	NAMED	LDHMAVIPBRSVRG-UHFFFAOYAE	C[N+]1=CC(C([NH-])=O)=CC=C1	1005-24-9	8305504	HMDB0000699	HMDB0000699	C02918	457	457

Given the identifier mappings extracted from CHEBI & Swisslipids, how many terms from this table can we map to an ontology identifier (=> separate breakdowns for CHEBI & Swisslipids).

If many do not map, I would be interested to discuss with chemi-informatics experts whether there are good strategies for mapping up. I'm hoping it would be enough to have a short discussion on OBO Slack with some examples.

Motivating issue: EBISPOT/efo#1905

Note - Santhi originally requested 415 terms (although the abstract mentions associations for 690 metabolites). Following discussion of identifiers, she has reduced her list to just 28 terms, but I think that is just because we communicated back that IDs are needed and she chose a subset of possible IDs (CAS ID/KEGG ID/PubChem ID), not because the science or needs of the GWAS catalog dictate adding such a small set.

[cmungall]

related:

• coordinate with SIB on conversion of swisslipids biopragmatics/pyobo#140