Incorrect Labels for some UniProtKB proteins in KG2

[isbluis]

A few dozen protein entries from UniProtKB appear to have incorrect labels, seen as having the string "Synonyms=xxxx" appended to them, e.g.: https://arax.ncats.io/devLM/index.html?term=UniProtKB:P55316

Full list:

Entry	Label	Type
UniProtKB:A5PKW4	PSDSynonyms=EFA6,	protein
UniProtKB:O14493	CLDN4Synonyms=CPER	protein
UniProtKB:O15047	SETD1ASynonyms=KIAA0339	protein
UniProtKB:O15523	DDX3YSynonyms=DBY	protein
UniProtKB:O43504	LAMTOR5Synonyms=HBXIP	protein
UniProtKB:O60231	DHX16Synonyms=DBP2,	protein
UniProtKB:O94759	TRPM2Synonyms=EREG1	protein
UniProtKB:O95259	KCNH1Synonyms=EAG	protein
UniProtKB:P00403	MT-CO2Synonyms=COII,	protein
UniProtKB:P04629	NTRK1Synonyms=MTC,	protein
UniProtKB:P32418	SLC8A1Synonyms=CNC,	protein
UniProtKB:P43003	SLC1A3Synonyms=EAAT1	protein
UniProtKB:P51648	ALDH3A2Synonyms=ALDH10	protein
UniProtKB:P55316	FOXG1Synonyms=FKH2,	protein
UniProtKB:P55735	SEC13Synonyms=D3S1231E,	protein
UniProtKB:P59901	LILRA4Synonyms=ILT7	protein
UniProtKB:P60014	KRTAP10-10Synonyms=KAP10.10,	protein
UniProtKB:P60412	KRTAP10-11Synonyms=KAP10.11,	protein
UniProtKB:P60413	KRTAP10-12Synonyms=KAP10.12,	protein
UniProtKB:Q01167	FOXK2Synonyms=ILF	protein
UniProtKB:Q01650	SLC7A5Synonyms=CD98LC,	protein
UniProtKB:Q06416	POU5F1BSynonyms=OCT4PG1,	protein
UniProtKB:Q06432	CACNG1Synonyms=CACNLG	protein
UniProtKB:Q0WX57	USP17L24Synonyms=USP17,	protein
UniProtKB:Q13573	SNW1Synonyms=SKIIP,	protein
UniProtKB:Q13698	CACNA1SSynonyms=CACH1,	protein
UniProtKB:Q14494	NFE2L1Synonyms=HBZ17,	protein
UniProtKB:Q15648	MED1Synonyms=ARC205,	protein
UniProtKB:Q15811	ITSN1Synonyms=ITSN	protein
UniProtKB:Q15848	ADIPOQSynonyms=ACDC,	protein
UniProtKB:Q2NKX8	ERCC6LSynonyms=PICH	protein
UniProtKB:Q3V6T2	CCDC88ASynonyms=APE	protein
UniProtKB:Q5W186	CST9Synonyms=CLM	protein
UniProtKB:Q5XXA6	ANO1Synonyms=DOG1	protein
UniProtKB:Q6ZR08	DNAH12Synonyms=DHC3,	protein
UniProtKB:Q7L5Y9	MAEASynonyms=EMP	protein
UniProtKB:Q7Z6J0	SH3RF1Synonyms=KIAA1494,	protein
UniProtKB:Q86SK9	SCD5Synonyms=ACOD4	protein
UniProtKB:Q8IUG1	KRTAP1-3Synonyms=B2B,	protein
UniProtKB:Q8IWB7	WDFY1Synonyms=FENS1	protein
UniProtKB:Q8IWT6	LRRC8ASynonyms=KIAA1437,	protein
UniProtKB:Q8IXQ6	PARP9Synonyms=BAL	protein
UniProtKB:Q8IXZ2	ZC3H3Synonyms=KIAA0150,	protein
UniProtKB:Q8IZJ1	UNC5BSynonyms=P53RDL1	protein
UniProtKB:Q8NDX1	PSD4Synonyms=EFA6B	protein
UniProtKB:Q8NHM5	KDM2BSynonyms=CXXC2,	protein
UniProtKB:Q8TEC5	SH3RF2Synonyms=POSH3	protein
UniProtKB:Q92614	MYO18ASynonyms=CD245	protein
UniProtKB:Q92990	GLMNSynonyms=FAP48	protein
UniProtKB:Q96BI1	SLC22A18Synonyms=BWR1A,	protein
UniProtKB:Q96GG9	DCUN1D1Synonyms=DCN1	protein
UniProtKB:Q96J94	PIWIL1Synonyms=HIWI	protein
UniProtKB:Q96KP4	CNDP2Synonyms=CN2,	protein
UniProtKB:Q96KV7	WDR90Synonyms=C16orf15,	protein
UniProtKB:Q96PE5	OPALINSynonyms=HTMP10	protein
UniProtKB:Q9BSY9	DESI2Synonyms=C1orf121,	protein
UniProtKB:Q9BUR5	APOOSynonyms=FAM121B,	protein
UniProtKB:Q9BYW2	SETD2Synonyms=HIF1,	protein
UniProtKB:Q9C000	NLRP1Synonyms=CARD7,	protein
UniProtKB:Q9H4E5	RHOJSynonyms=ARHJ,	protein
UniProtKB:Q9HD42	CHMP1ASynonyms=CHMP1	protein
UniProtKB:Q9NTI5	PDS5BSynonyms=APRIN,	protein
UniProtKB:Q9NTJ5	SACM1LSynonyms=KIAA0851	protein
UniProtKB:Q9NWU2	GID8Synonyms=C20orf11,	protein
UniProtKB:Q9NWV8	BABAM1Synonyms=C19orf62,	protein
UniProtKB:Q9P0L9	PKD2L1Synonyms=PKD2L,	protein
UniProtKB:Q9UBG3	CRNNSynonyms=C1orf10	protein
UniProtKB:Q9UPN3	MACF1Synonyms=ABP620,	protein
UniProtKB:Q9Y2K7	KDM2ASynonyms=CXXC8,	protein
UniProtKB:Q9Y4F9	RIPOR2Synonyms=C6orf32,	protein

[saramsey]

Hi @isbluis thank you for bringing this to our attention.

This appears to be a bug in KG2. Verified the bug in KG2.5.1 using cypher

match (n {id: 'UniProtKB:A5PKW4'}) return n.id, n.name, n.full_name;

results show a weird name field consistent with what was seen in the UI. In KG2 we have:

[isbluis]

Great, thanks for looking into it @saramsey ! (and sorry if I abused the tag; was not sure which to use)

[kvarforl]

Okay I'm making progress on this issue, and in doing so I noticed a small bug that was keeping the GN 'synonyms' from being appended to the node synonyms. I fixed this, but now some of the synonyms have evidence codes attached to them.

I'm assuming this is not desirable, so I'm going to remove them for now, but note it here for if we ever want to do something specific with them.

As an example, here are the GN lines of uniprot_dat file entry for UniProtKB:Q9Y4F9.

GN   Name=RIPOR2;
GN   Synonyms=C6orf32, DIFF48, FAM65B, KIAA0386,
GN   PL48 {ECO:0000303|PubMed:9055809};

[isbluis]

Great, thanks for fixing this, @kvarforl !

One idea that recently occurred to me is that it might be worth looking at the Uniprot entries in KG2 that have the longest string values for name (say, the top 20 or 50) as a way to see if there are other potential parsing bugs still lurking -- most protein (short) names being only 4-6 characters long. Perhaps you are already doing this or something better, so apologies if this is not too useful.

[kvarforl]

@isbluis this is a great idea! I've recently been pondering various ways to catch some of the kg2 bugs before someone has to stumble upon and discover them, but haven't gotten around to doing any of it. This sounds like a great place to start. thanks for the suggestion!

[isbluis]

Excellent! Perhaps another way is to look for strings in that same field that contain characters that are non-alpha/numerical (e.g. equals sign, comma, etc.) Perhaps even lowercase letters?

[kvarforl]

Fixed in kg2.5.2