The pipeline behind the GO:0016020 row. Its input for ADCK5 is SubCell SL-0162 (Membrane), which resolves to GO:0016020 and nothing more specific.

Source of the first GO:0005515 row. Reports an ADCK5-NOTCH2NLA interaction by yeast two-hybrid, logged in IntAct three times under three sub-method labels of the same screen.

Source of the second GO:0005515 row; recovers the same ADCK5-NOTCH2NLA pair by two hybrid array from the same CCSB resource lineage, so it is method replication rather than orthogonal validation.

MitoCoP places ADCK5 in the high-confidence human mitochondrial proteome; this is the source of the GO:0005739 HTP row, and the identical row is carried by ADCK1 and ADCK2.

Defines the UbiB-specific sequence features, including the invariant KxGQ motif and the AAAS alanine-rich loop that replaces the canonical glycine-rich nucleotide-binding loop. Both are retained in ADCK5, at K147 and A209.

Argues that the KxGQ domain is positioned to inhibit protein kinase activity by occupying the site conventional kinases use for peptide substrates.

Establishes that eukaryotic UbiB proteins are mitochondrial, corroborating ADCK5's HTP mitochondrion row independently of the proteomics.

Demonstrates that COQ8A and yeast Coq8p lack canonical protein kinase activity in trans; the source of the NOT|enables GO:0004672 and NOT|involved_in GO:0006468 annotations carried by both COQ8A and COQ8B.

Attributes the loss of generic protein-kinase function to the KxGQ domain blocking the peptide substrate site.

Explicitly leaves ADCK1, ADCK2 and ADCK5 uncharacterised, so nothing in this paper is direct evidence about ADCK5.

COQ8B phosphorylates a specific protein substrate, COQ3, at multiple sites - the result that stops this review concluding that the UbiB fold is incapable of phosphorylating a protein.

The same study excludes small-molecule kinase activity for COQ8B, narrowing the family's plausible activities.

The subcellular kinome atlas underpins the experimental mitochondrial localisation (ECO:0000269|PubMed:33988507) that UniProt gives every human UbiB protein except ADCK5 - though not to the same degree, and the difference matters. For ADCK1 and ADCK2 it is the SOLE cited evidence, so there it supplied the annotation. For COQ8A and COQ8B it is one tag among several and the latest of them (COQ8A also cites PubMed:11888884 from 2002 and PubMed:25498144; COQ8B also cites PubMed:24270420 from 2013), so there it corroborated an annotation that already existed. ADCK5 has no tag from it at all, because it states that ADCK5 was not present in the screened library - so ADCK5's missing annotation reflects absence of testing rather than a negative result, which is what makes the UniProt correction request reasonable rather than a request to overrule data. The four genes the screen did cover are the entire rest of the human UbiB family; ADCK5 is the only member it could not assess. The supplied/corroborated split is computed in family_census.json under mitochondrial_localisation_provenance, not asserted from memory.

Stronger than a bare absence. The sentence sits in a passage reconciling the screen's 22 mitochondrial kinases against the MitoCarta2.0 mitochondrial proteome, and that passage lists two distinct kinds of exception. ADCK5 and OBSCN are in the first kind: absent from the library, so never assessed. FASTK and PAK5 are in the second and opposite kind: they were in the library, were tested, and were called non-mitochondrial. ADCK5 belongs to the first group only - a mitochondrial candidate the assay never reached, not one it examined and rejected.

Contributes 40 of ADCK5's 54 IntAct records, yielding 25 distinct partners of which 17 are UniProt-annotated to the mitochondrion. Independent corroboration of the mitochondrion row, and the source of the compartment consideration that sits alongside - but is not required by - the method-replication argument against the NOTCH2NLA pairing.

The only ADCK5-specific functional study. Proposes that ADCK5 phosphorylates SOX9 at Ser181 to upregulate PTTG1 and drive invasion and migration, but states the central claim as a possibility rather than a demonstration.

The Ser181 assignment rests on mutagenesis of the substrate plus a motif-match argument, not on a direct kinase assay.

MAFFT alignment of the five human UbiB proteins, yeast Coq8p, E. coli UbiB and PKA Calpha shows ADCK5 retains both UbiB inhibitory features (KxGQ K147; A-rich loop A209) and the complete catalytic core (K228, E281, D360, N365, D382). Published reference residues are asserted before use, and the two independently anchored A-rich-loop columns resolve to the same alignment column.

Census of the five human UbiB genes: ADCK5 is the only one with zero IBA annotations, and the EC downgrade to 2.7.-.- with loss of the Ser/Thr-kinase keyword has been applied only to COQ8A and COQ8B, the two members actually assayed.

All 54 IntAct records for Q3MIX3 were retrieved without truncation; the NOTCH2NLA interaction is one Y2H screen logged under three sub-method labels plus one further Y2H from the same resource lineage, with no orthogonal assay.