Which KP can provide metabolic pathway data for Translator?

[karafecho]

The QotM identified PharmGKB, RaMP DB, Rhea-DB, and KEGG Pathway DB as potential sources of metabolic pathway data.

[marcdubybroad]

@karafecho The Genetics KP team is in the process of promoting new data through our stack that has KEGG and WIKIPATHWAYS data. It's currently in our dev stack.
We've only loaded disease/pathway association data where the p-value is less than 0.05.
The p-values are calculated from the Magma process aggregation of the genes' p-values in the pathway for that disease/phenotype.
Example query for our dev server (https://translator.broadinstitute.org/genetics_provider/trapi/v1.3/query)

{
    "message": {
      "query_graph": {
        "nodes": {
          "n0": {
            "ids": ["KEGG:hsa05322"]
          },
          "n1": {
            "categories": ["biolink:NamedThing"]
          }
        },
        "edges": {
          "e01": {
            "subject": "n0",
            "object": "n1"
          }
        }
      }
    }
  }

returns:

                        {
                            "attribute_type_id": "biolink:p_value",
                            "original_attribute_name": "pValue",
                            "value": 0.0019390149921821016
                        }
                    ],
                    "object": "EFO:0004308",
                    "predicate": "biolink:genetic_association",
                    "subject": "KEGG:hsa05322"
                }

[karafecho]

Thanks, Mark! This type of query could be very useful for Translator.

[sstemann]

@sierra-moxon @karafecho NCATS will be creating an RaMP DB KP over the summer

[karafecho]

Wow! Thanks for the update, Sarah.

[andrewsu]

@sstemann Can you put us in touch with the developers of the RaMP API? Since the API already exists, it's possible we can annotate/register it in SmartAPI and have it accessible through BTE without any further development. We were about to get on that, but have one detail we'd like to follow up on first.

[sandrine-m]

This is linked with #2

[sierra-moxon]

@andrewsu @sstemann - did either of these resources get incorporated into a Translator resource or should we keep this ticket open for further work?

[andrewsu]

Service Provider provides access to a rhea API -- more info in biothings/mychem.info#37

RaMP is ongoing -- linking the currently open issue at biothings/biothings_explorer#705

[codewarrior2000]

MolePro is in the process of assimilating the Reactome DB, which is "an open-source, open access, manually curated and peer-reviewed pathway database".

In Reactome, the metabolic pathways are grouped under Metabolism, which has the Reactome identifier R-HSA-1430728.

[sstemann]

we deferred to BTEs RaMP integration. If any of these are targeted for this funding phase of the program can folks please add a release (guppy, hammerhead, etc) to the label? @codewarrior2000 @andrewsu

[andrewsu]

Given the recent emphasis on eliminating redundant ingestion of resources into Translator, I'm rethinking how we approach RaMP. TLDR: I don't think we should ingest RaMP directly, but rather focus on Lipid Maps, the only resource that isn't otherwise already present in Translator. More details below...

According to https://rampdb.nih.gov/about, RAMP is an aggregator of the following resources:

HMDB v5.0 (2021-11-17)
Reactome v87 (Dec 2023)
WikiPathways v20240210 (2024-02-10)
KEGG from HMDB (v5.0) (2021-11-17)
ChEBI Release 231 (2024-03-01)
Lipid Maps Release 2024-03-06
Rhea Release 131 (2024-01-)

My analysis of each one of those resources:

HMDB

• available from Service Provider at https://biothings.transltr.io/hmdb, but not queried by BTE
• also available through Automat-hmdb(Trapi v1.5.0) at https://smart-api.info/registry?q=0a1c0f46f4950b82b1aa7dad27aad10a queried by BTE

Reactome

• available through mygene.info smartAPI annotation, queried by BTE
• also through Automat-reactome(Trapi v1.5.0) http://smart-api.info/registry?q=61b41c5d9b90eb8ad16e037f9a87d593

Wikipathways

• available through mygene.info smartAPI annotation, queried by BTE

KEGG

• available through mygene.info smartAPI annotation, queried by BTE

ChEBI

• available through mychem.info smartAPI annotation, queried by BTE

Lipid Maps

• not in infores catalog, not being queried
• available in mychem.info (metadata fields), so in principle can be added

Rhea

• available from Service Provider at https://biothings.transltr.io/rhea, queried by BTE

Bottom line, the only one that isn't currently ingested into Translator is Lipid Maps. Unless there are objections, I think we should:

• create an entry for Lipid Maps in the infores catalog
• expand the mychem.info smartAPI annotation to include LipidMaps

There also is the issue of whether the resources above have been captured as completely in the sources above as they have been in RaMP, but barring a drastic difference in completeness/accuracy, I think we should focus on improving the existing resources over adding new ones.

[Mathelab]

@andrewsu I agree that it makes more sense to improve upon the existing resources. A couple of nuances worth mentioning:

• RaMP does 'harmonize' metabolite names across the resources, which is actually very difficult, particularly for lipids.
• Biochemical pathways and chemistry seem to be pulled into Translator from other sources listed above so those may not be useful. Not sure whether Classyfire is being pulled in but that ontology is very widely used in the metabo community.
• Seems like you could pull from LIPID MAPS directly or through RaMP to take advantage of the 'harmonization' and our parsing code (which is also public).
• RaMP pulls in ontologies from HMDB not sure whether those would be useful.

Happy to discuss more during the relay as well.