ZNF788P notes

2026-10-04 Tier 3 over-annotation audit (claude-code)

UniProt Q6ZQV5 (ZN788_HUMAN), "Putative KRAB domain-containing protein ZNF788", 82 aa,
PE5 (uncertain), CAUTION "Could be the product of a pseudogene". Both FLJ cDNAs
(BAC87366/AK128282 = FLJ46419, BAC87578/AK128700) are flagged "Wrong choice of CDS".
Keyword "Proteomics identification" (PeptideAtlas/MassIVE cross-refs); I did not
determine which peptides, or whether they are unique given the highly similar chr19
ZNF paralogs.
HGNC (REST, 2026-10-04): HGNC:33112, "zinc finger family member 788, pseudogene",
locus_type pseudogene, 19p13.2, previous symbol ZNF788, alias FLJ46419.
Ensembl ENSG00000214189: transcribed_unprocessed_pseudogene; the single transcript has
no translation.
Literature: PubMed ZNF788P/ZNF788 returns only methylation-marker papers (anal
cancer, colorectal cancer, eGFR in HIV, kidney transplant). These are CpG associations
at the locus, not protein function.

Sequence versions 1-2 (2004-2018): 615-aa protein with 14 C2H2 zinc fingers and no
KRAB. Version 3 (since 2018): 82-aa truncated KRAB-A with 0 zinc fingers.
KRAB-A aligns to ZNF10 KRAB 48% identity but ends at the ZNF10 equivalent of residue
There is no B box. DV is retained; MLE becomes MQE.
In AK128700 the KRAB exon and the zinc-finger ORF are in the same frame but separated
by a stop codon. The locus looks like a KZFP whose KRAB and zinc-finger parts are no
longer joined.

This was NOT an overexpression study. Abstract: PMID:16791210.
Full text (author-hosted PDF, Carpenter-Singh lab site
https://carpenter-singh-lab.broadinstitute.org/files/anne/files/20-Sigal_NatMethods_2006.pdf;
not in the publications cache, which is abstract-only): CD-tagging inserts a
promoterless, start-codon-less YFP exon into introns in H1299 cells, and "The
YFP-tagged protein in each clone was identified by 3' RACE". Fig. 3 panel (q) is
"uncharacterized protein FLJ46419", one of the 20 nuclear proteins studied. FLJ46419
is the HGNC alias of ZNF788P.
Implication: a YFP fusion expressed from this locus (or from a transcript that 3' RACE
assigned to it) is nuclear in H1299 cells. YFP has no ATG, so the upstream exons must
be translated in frame, and 3' RACE reads the downstream sequence. The tagged product
therefore most likely includes downstream (zinc-finger) sequence, not the 82-aa KRAB
fragment. The GOA row is dated 2015-01-22, when the entry was the 615-aa zinc-finger
sequence (version 2).
Conflict: real endogenous evidence that the locus makes a nuclear protein, versus
HGNC/Ensembl pseudogene status and the 2018 UniProt remodelling to an 82-aa fragment.
I cannot tell from here whether the 3' RACE read was unique to ZNF788P rather than a
paralogous chr19 zinc-finger gene. -> UNDECIDED.

KRAB repression is DNA-binding-dependent: PMID:8183939. The 82-aa product has no DNA-binding domain and
comes from an annotated pseudogene. -> REMOVE.

GO:0006355 comes from InterPro2GO on the KRAB signature (IPR001909, IPR036051). The
mapping triggers on any KRAB-A match, including truncated fragments with no zinc
fingers.
GO:0005634 is HDA, assigned by UniProt (2015) against the former 615-aa sequence and
carried over when the sequence was replaced in 2018.