bridgedb / Goliath

0 stars 1 forks source link

Metabolite ID conversion gives different results, based on auto-select yes/no #7

Open DeniseSl22 opened 5 years ago

DeniseSl22 commented 5 years ago

Hi @nunogit ;

I'm testing goliath with the list of HMDb IDs that Freddie was working on: HMDB00112 HMDB00295 HMDB00901 HMDB00935 HMDB01275 HMDB01877 HMDB13037

When I do autoselect, I get these results: image

When I say I want to convert from HMDb to CHEBI, the results change (I have no mappings left): image

And, when I change the IDs of HMDb from old to new ID structure, I get the following results: image

(every ID but the last is similar... this could be due to mapping issues).

And with HMDB selected as primary database to map from (some mappings, but most do not work): image

And last but not least, the ID that keeps giving "undefined" as a result, does exist in HMDB: http://www.hmdb.ca/metabolites/HMDB0000901 . I can check if it is also part of our mapping files (could be that the webservice is still running on quite old data right?)...

nunogit commented 5 years ago

And last but not least, the ID that keeps giving "undefined" as a result, does exist in HMDB: http://www.hmdb.ca/metabolites/HMDB0000901 . I can check if it is also part of our mapping files (could be that the webservice is still running on quite old data right?)...

Undefined does not necessarly mean the id as a source doesn't exist; It means that there is no target id; But you raise a good point; We should make these 2 situations clear.

DeniseSl22 commented 5 years ago

Yes true, apparently neither HMDb nor Wikidata (https://www.wikidata.org/wiki/Q47538360) have a mapping to ChEBI, so then it probably does not exist in ChEBI (yet)... So if there is no target ID, the word undefined could be replaced by "no mapping available"?

DeniseSl22 commented 5 years ago

So I think the autoselect vs selecting a database to map from difference is a bug.... The other one (secondary and primary IDs) should be dealt with at some point (but not now, since updates from BridgeDb are needed for this).