Found document where drugcentral value is array, but items are for different chemicals

[colleenXu]

As part of biothings/biothings.api#368, I'm reviewing documents where drugcentral is an array to see this data-structure makes sense.

I've come across an example where one of the drugcentral items looks like it's in the wrong document: https://mychem.info/v1/chem/CPZBLNMUGSZIPR-NVXWUHKLSA-N

• the rest of the data in the document is for palonosetron
• But the 2nd item in the drugcentral array is for aurothioglucose, which is a completely different chemical

[colleenXu]

I'm not sure about these, because they're kinda similar but also kinda not:

• https://mychem.info/v1/chem/IZTQOLKUZKXIRV-YRVFCXMDSA-N: the rest of the data in the document is for Sincalide. But the 1st item in the drugcentral array is for pancreozymin / cholecystokinin, which seems to be a more general or longer version of the peptide. So they're kinda different. But they do seem to have the same purpose.
• https://mychem.info/v1/chem/OMPJBNCRMGITSC-UHFFFAOYSA-N: the rest of the data in the document is for benzoyl peroxide. But the 2nd item in the drugcentral array is for "benzamycin" which is a drug that's a mix of benzoyl peroxide and erythromycin.
• https://mychem.info/v1/chem/ULGZDMOVFRHVEP-RWJQBGPGSA-N: the rest of the data in the document is for erythromycin. But the 1st item in the drugcentral array is for benzamycin (same situation as benzoyl peroxide?)
• https://mychem.info/v1/chem/WFWLQNSHRPWKFK-UHFFFAOYSA-N: the rest of the data in the document is for tegafur. But the 1st item in the drugcentral array is for "tegafur-uracil" which is a drug that's a mix of those two chemicals.

[colleenXu]

Here's an example of downstream effects (BTE querying and processing responses):

Take https://mychem.info/v1/chem/IZTQOLKUZKXIRV-YRVFCXMDSA-N. It has two drugcentral items with slightly diff bioactivity contents (pointing out 1 diff here):

• cholecystokinin (drugcentral.xrefs.umls ==C0008328). Only this item has bioactivity objects with action_type = AGONIST: uniprot IDs are P32239 and P30551
• sincalide (drugcentral.xrefs.umls ==C0037167)

If I start from the chemical ID sincalide (C0037167) and action_type=AGONIST, BTE will send a query like this and cholecystokinin (C0008328)'s data will be returned (it should return nothing for sincalide C0037167 alone). So BTE will link sincalide with the returned uniprot IDs (create edges), which isn't really desired behavior. But it's understandable because the 2 chemicals' drugcentral data are together in 1 document.

Query from chemical ID for sincalide

curl --location --globoff 'https://mychem.info/v1/query?fields=drugcentral.bioactivity&size=1000&with_total=true&jmespath=drugcentral.bioactivity%7C[%3Faction_type%3D%3D%60AGONIST%60%20%7C%7C%20action_type%3D%3D%60PARTIAL%20AGONIST%60]&always_list=drugcentral.bioactivity&filter=drugcentral.bioactivity.action_type%3A(%22AGONIST%22%20OR%20%22PARTIAL%20AGONIST%22)' \
--header 'Content-Type: application/json' \
--data '{
    "q": ["C0037167"],
    "scopes": "drugcentral.xrefs.umlscui"
}'

Note: The query jmespath could be modified to add the chemical umls ID. But I kinda don't want to do that because then this type of query wouldn't be a batch-query anymore. And it kinda covers up this data problem...

[colleenXu]

FYI this problem doesn't happen if I query from the "reverse" direction.

If I start from the uniprot ID (ex: P32239 for action_type=AGONIST), BTE will send a query like this and only cholecystokinin (C0008328)'s data will be returned (not sincalide C0037167).

Query from UniProt ID

curl --location --globoff 'https://mychem.info/v1/query?size=1000&fields=drugcentral.bioactivity%2Cdrugcentral.xrefs.umlscui&jmespath_exclude_empty=true&always_list=drugcentral.bioactivity&jmespath=drugcentral.bioactivity%7C[%3F(action_type%3D%3D%60AGONIST%60%20%7C%7C%20action_type%3D%3D%60PARTIAL%20AGONIST%60)%20%26%26%20length(uniprot[%3Funiprot_id%3D%3D%27P32239%27])%20%3E%20%600%60]&filter=drugcentral.bioactivity.action_type%3A(%22AGONIST%22%20OR%20%22PARTIAL%20AGONIST%22)%20AND%20_exists_%3Adrugcentral.xrefs.umlscui' \
--header 'Content-Type: application/json' \
--data '{
    "q": ["P32239"],
    "scopes": "drugcentral.bioactivity.uniprot.uniprot_id"
}'