Issues with BioThings BindingDB object fields

[colleenXu]

(CC @newgene @erikyao for pending BioThings, @rjawesome as the original person who worked on the parser biothings/pending.api#70)

Andy Crouse from Translator's UI team pointed out that BTE was returning result chemical Nodes that didn't have names, with edges from BioThings BindingDB (Translator Slack link).

When investigating this, I discovered cases where object fields (the chemical), specifically ID and name ones, seem incorrect or problematic (see the comments). Some notes:

• I've been assuming that the object.pubchem_sid value is a "correct" ID (when the document has this field, which is >98% of documents or 1413131 / 1438909)
• I'm not sure how many documents are affected
• I'm not sure if the other chemical (object) fields have similar issues

[colleenXu]

object.pubchem_cid

• seems tricky to fix (or not fixable, issues are upstream in BindingDB data source?)
• values sometimes seem incorrect

expand for detailed example

Example: https://biothings.ncats.io/bindingdb/query?q=object.pubchem_cid:4585

"object":{
                "bindingdb_link": "http://www.bindingdb.org/bind/chemsearch/marvin/MolStructure.jsp?monomerid=35254",
                "chebi": "7735",
                "chembl": "CHEMBL715",
                "drugbank": "DB00334",
                "inchi": "InChI=1S/C17H20N4S/c1-12-11-13-16(21-9-7-20(2)8-10-21)18-14-5-3-4-6-15(14)19-17(13)22-12/h3-6,11,19H,7-10H2,1-2H3",
                "inchikey": "KVWDHTXUZHCGIO-UHFFFAOYSA-N",
                "iuphar_grac_id": "47",
                "kegg": "C07322",
                "monomer_id": 35254,
                "name": "2-methyl-4-(4-methylpiperazin-1-yl)-10H-thieno[2,3-b][1,5]benzodiazepine::CHEMBL715::OLANZAPINE::Olansek::US8802672, Olanzapine::Zyprexa::olanzapine",
                "pubchem_cid": 4585,
                "pubchem_sid": 85753333,
                "smiles": "CN1CCN(CC1)C1=Nc2ccccc2Nc2sc(C)cc12",
                "zinc": "ZINC52957434"
            },

• pubchem compound ID (CID) 4585 doesn't exist: https://pubchem.ncbi.nlm.nih.gov/#query=4585
• pubchem substance ID (SID) 85753333 does exist. It links to the pubchem cid 135398745 (Olanzapine)
• I checked the chebi, chembl, drugbank, inchikey, and kegg IDs; these are all mapped IDs on the pubchem cid 135398745 (Olanzapine) page
• could 4585 be from the ID NSC_4585? I see that ID as a "synonym" on the pubchem cid 135398745 (Olanzapine) page

Other examples with basically the same behavior (notes here):

• 53322247
• 50908790
• 53326169
• 2818

[colleenXu]

object.inchikey

• seems tricky to fix (or not fixable, issues are upstream in BindingDB data source?)
• not clear if these values are incorrect, outdated, or represent slightly different but valid structures (isomers? different salts or protonation states?)...

example 1

• from https://biothings.ncats.io/bindingdb/query?q=object.inchikey:YQCLAYRIYWYIKH-UHFFFAOYSA-N, see code chunk below for the object
• the pubchem cid (9579327)'s page has a different inchikey: YQCLAYRIYWYIKH-WGPBWIAQSA-N
  • Translator's Node Norm doesn't recognize BindingDB's inchikey but does recognize the pubchem cid's inchikey
• the issue could be that there's multiple isomers/chemicals involved? Notice how there's two chembl IDs in object.chembl and object.name fields...

"object": {
                "bindingdb_link": "http://www.bindingdb.org/bind/chemsearch/marvin/MolStructure.jsp?monomerid=50115522",
                "chembl": "CHEMBL341945CHEMBL106813",
                "inchi": "InChI=1S/C28H35F3N4O4/c1-19-9-16-35(37)20(2)24(19)26(36)33-17-12-27(3,13-18-33)34-14-10-22(11-15-34)25(32-38-4)21-5-7-23(8-6-21)39-28(29,30)31/h5-9,16,22H,10-15,17-18H2,1-4H3",
                "inchikey": "YQCLAYRIYWYIKH-UHFFFAOYSA-N",
                "monomer_id": 50115522,
                "name": "(2,4-Dimethyl-1-oxy-pyridin-3-yl)-{4-[methoxyimino-(4-trifluoromethoxy-phenyl)-methyl]-4'-methyl-[1,4']bipiperidinyl-1'-yl}-methanone::CHEMBL106813::CHEMBL341945",
                "pubchem_cid": 9579327,
                "pubchem_sid": 104009437,
                "smiles": "CO[N-][C+](C1CCN(CC1)C1(C)CCN(CC1)C(=O)c1c(C)cc[n+]([O-])c1C)c1ccc(OC(F)(F)F)cc1",
                "zinc": "ZINC26848826"
            },

example 2

• https://biothings.ncats.io/bindingdb/query?q=object.inchikey:ZUXABONWMNSFBN-UHFFFAOYSA-N, see code chunk below for the object. This object ALSO has the "incorrect" object.pubchem_cid field issue (listed in the previous comment as pubchem_cid 2818)
• pubchem compound ID (CID) 2818 doesn't exist: https://pubchem.ncbi.nlm.nih.gov/#query=2818
• pubchem substance ID (SID) 49846683 does exist. It links to the pubchem cid 135398737 (Clozapine)
• but pubchem cid 135398737 (Clozapine)'s page has a different inchikey: QZUDBNBUXVUHMW-UHFFFAOYSA-N
  • Translator's Node Norm seems to recognize BindingDB's inchikey and pubchem cid's inchikey as two separate entities
• interestingly, BindingDB's inchikey does show up in a source url on pubchem cid 135398737 (Clozapine)'s page (search for "MassBank of North America (MoNA)" and look at the urls)

"object": {
                "bindingdb_link": "http://www.bindingdb.org/bind/chemsearch/marvin/MolStructure.jsp?monomerid=22869",
                "inchi": "InChI=1S/C18H19ClN4/c1-22-8-10-23(11-9-22)18-14-4-2-3-5-15(14)20-16-7-6-13(19)12-17(16)21-18/h2-7,12,21H,8-11H2,1H3",
                "inchikey": "ZUXABONWMNSFBN-UHFFFAOYSA-N",
                "iuphar_grac_id": "38",
                "monomer_id": 22869,
                "name": "6-chloro-10-(4-methylpiperazin-1-yl)-2,9-diazatricyclo[9.4.0.0^{3,8}]pentadeca-1,3(8),4,6,10,12,14-heptaene::CLOZARIL::Clozapine::Leponex::US10259786, Clozapine",
                "pubchem_cid": 2818,
                "pubchem_sid": 49846683,
                "smiles": "CN1CCN(CC1)C1=c2ccccc2=Nc2ccc(Cl)cc2N1",
                "zinc": "ZINC19796155"
            },

[colleenXu]

object.chembl

Sometimes the values seem to be concatenated strings of multiple IDs:

• seems EASY to fix: split whenever CHEMBL shows up more than once (clue that there's > 1 CHEMBL ID in the string). have an array of strings where each string is a single CHEMBL ID
• I haven't done any checking to see if the single CHEMBL IDs are valid or not

Examples:

• CHEMBL343826CHEMBL419735: seems to be a parsing issue. Are these two separate chemicals (CHEMBL343826 and CHEMBL419735)?
• CHEMBL341945CHEMBL106813
• CHEMBL141345CHEMBL107999
• CHEMBL138263CHEMBL106558
• CHEMBL107373CHEMBL140283

[colleenXu]

object.name

• seems easy to fix
• These seem to be "::"-delimited strings, that are sometimes very long and unwieldy. I think it'd make sense to split them on "::" and have arrays of strings instead.

For examples, click on any of the BioThings BindingDB links above, and look at the object.name value.

[colleenXu]

And this is what I did to address Andy Crouse's original problem of "Nodes with no names from BioThings BindingDB":

(pasted from biothings/pending.api#99 (comment))

For now, I've changed the x-bte annotation for this resource NCATS-Tangerine/translator-api-registry@022e876:

(1) Use object.inchikey:

• covers a little over 96% of the resource (1394153 / 1438909)
• my hunch is that the INCHIKEY IDs are not completely incorrect, VS the pubchem cids are sometimes incorrect and I don't really know why
• VS object.pubchem_cid (used before) covers a little over 98% of the resource (1413051 / 1438909)
• other fields cover much less of the resource or aren't supported by Node Norm:
  • object.pubchem_sid: 1413131 but Node Norm doesn't seem to support this ID namespace right now
  • object.chembl: 631745. Has its own issues, see last comment
  • object.kegg (KEGG.COMPOUND): 33680
  • object.chebi: 25903
  • object.drugbank: 24177

(2) Retrieve subject.name and object.name fields for input_name/output_name behavior, if Node Norm doesn't retrieve info for the ID. Every document has those fields, but the names provided have issues (will be covered in a later post).

[colleenXu]

@everaldorodrigo @newgene @andrewsu

I think this would be a useful issue for @everaldorodrigo to dig into and address as much as possible.

[everaldorodrigo]

• I've been assuming that the object.pubchem_sid value is a "correct" ID (when the document has this field, which is >98% of documents or 1413131 / 1438909)
• I'm not sure how many documents are affected
• I'm not sure if the other chemical (object) fields have similar issues

@colleenXu, considering the last released data from May 2024 to the CI environment,
a total of 41678 from 1795611 items don't have the field object.pubchem_sid.

For those cases missing the object.pubchem_sid field, seems the data source (.tsv file) doesn't have the value for the column PubChem SID used in the parser to fill the field object.pubchem_sid.

The partial data below is extracted from the data source. There are two lines. The header and an example of item missing the field PubChem SID. Look for PubChem SID in the table below:

BindingDB Reactant_set_id	Ligand SMILES	Ligand InChI	Ligand InChI Key	BindingDB MonomerID	BindingDB Ligand Name	Target Name	Target Source Organism According to Curator or DataSource	Ki (nM)	IC50 (nM)	Kd (nM)	EC50 (nM)	kon (M-1-s-1)	koff (s-1)	pH	Temp (C)	Curation/DataSource	Article DOI	BindingDB Entry DOI	PMID	PubChem AID	Patent Number	Authors	Institution	Link to Ligand in BindingDB	Link to Target in BindingDB	Link to Ligand-Target Pair in BindingDB	Ligand HET ID in PDB	PDB ID(s) for Ligand-Target Complex	PubChem CID	PubChem SID	ChEBI ID of Ligand	ChEMBL ID of Ligand	DrugBank ID of Ligand	IUPHAR_GRAC ID of Ligand	KEGG ID of Ligand	ZINC ID of Ligand	Number of Protein Chains in Target (>1 implies a multichain complex)	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain	BindingDB Target Chain Sequence	PDB ID(s) of Target Chain	UniProt (SwissProt) Recommended Name of Target Chain	UniProt (SwissProt) Entry Name of Target Chain	UniProt (SwissProt) Primary ID of Target Chain	UniProt (SwissProt) Secondary ID(s) of Target Chain	UniProt (SwissProt) Alternative ID(s) of Target Chain	UniProt (TrEMBL) Submitted Name of Target Chain	UniProt (TrEMBL) Entry Name of Target Chain	UniProt (TrEMBL) Primary ID of Target Chain	UniProt (TrEMBL) Secondary ID(s) of Target Chain	UniProt (TrEMBL) Alternative ID(s) of Target Chain
1187404	CS(=O)(=O)N1CCC@HC@@HC1	InChI=1S/C20H27FN6O4S/c1-32(30,31)26-7-6-16(15(21)11-26)24-20-23-10-13-8-12(9-17(22)28)19(29)27(18(13)25-20)14-4-2-3-5-14/h8,10,14-16H,2-7,9,11H2,1H3,(H2,22,28)(H,23,24,25)/t15-,16-/m0/s1	LGDFLYMYRBSZOR-HOTGVXAUSA-N	370273	BDBM467168::US10233188, Example 160	Cyclin-dependent kinase 6/G1/S-specific cyclin-D1 [L188C]	Homo sapiens	3.55								US Patent		10.7270/Q2KK9G2V		aid1806677	US11396512	Behenna, DC; Chen, P; Freeman-Cook, KD; Hoffman, RL; Jalaie, M; Nagata, A; Nair, SK; Ninkovic, S; Ornelas, MA; Palmer, CL; Rui, EY	Pfizer Inc.	http://www.bindingdb.org/bind/chemsearch/marvin/MolStructure.jsp?monomerid=370273	http://www.bindingdb.org/jsp/dbsearch/PrimarySearch_ki.jsp?energyterm=kJ/mole&tag=com&complexid=99&target=Cyclin-dependent+kinase+6%2FG1%2FS-specific+cyclin-D1+%5BL188C%5D&column=ki&startPg=0&Increment=50&submit=Search	http://www.bindingdb.org/jsp/dbsearch/PrimarySearch_ki.jsp?energyterm=kJ/mole&tag=r21&monomerid=370273&enzyme=Cyclin-dependent+kinase+6%2FG1%2FS-specific+cyclin-D1+%5BL188C%5D&column=ki&startPg=0&Increment=50&submit=Search											2	mehqllccevetirraypdanllndrvlramlkaeetcapsvsyfkcvqkevlpsmrkivatwmlevceeqkceeevfplamnyldrflslepvkksrlqllgatcmfvaskmketipltaeklciytdgsirpeellqmelllvnklkwnlaamtphdfiehflskmpeaeenkqiirkhaqtfvascatdvkfisnppsmvaagsvvaavqglnlrspnnflsyyrltrflsrvikcdpdclracqeqieallesslrqaqqnmdpkaaeeeeeeeeevdlactptdvrdvdi		G1/S-specific cyclin-D1	CCND1_HUMAN	P24385	Q6LEF0							MEKDGLCRADQQYECVAEIGEGAYGKVFKARDLKNGGRFVALKRVRVQTGEEGMPLSTIREVAVLRHLETFEHPNVVRLFDVCTVSRTDRETKLTLVFEHVDQDLTTYLDKVPEPGVPTETIKDMMFQLLRGLDFLHSHRVVHRDLKPQNILVTSSGQIKLADFGLARIYSFQMALTSVVVTLWYRAPEVLLQSSYATPVDLWSVGCIFAEMFRRKPLFRGSSDVDQLGKILDVIGLPGEEDWPRDVALPRQAFHSKSAQPIEKFVTDIDELGKDLLLKCLTFNPAKRISAYSALSHPYFQDLERCKENLDSHLPPSQNTSELNTA	1G3N,1JOW,1XO2,2EUF,2F2C,3NUP,3NUX,4AUA,4EZ5,4TTH,5L2I,5L2S,5L2T,6OQL,6OQO	Cyclin-dependent kinase 6	CDK6_HUMAN	Q00534	A4D1G0

Do you think we should use another field for the operations instead of object.pubchem_sid?

[colleenXu]

Sorry for the late response.

I think it's okay that some rows don't have a pubchem SID. I've been using the INCHIKEY instead (see the earlier post where I explain that I think it covered most of the resource and was somewhat reliable but still had a problem).