ABCE1 (P61221) — sequence-level bioinformatics analysis

Reproducible support for three curation decisions in ABCE1-ai-review.yaml.
All numbers below are produced by analyze.py, which parses the curated
UniProt record ../ABCE1-uniprot.txt at run time. Nothing is hardcoded.

How to reproduce

cd genes/human/ABCE1/ABCE1-bioinformatics
python3 analyze.py            # pure standard library, no dependencies
# or, for a pinned environment:
uv run python analyze.py

Methods

Input: UniProt/Swiss-Prot record P61221 (ABCE1-uniprot.txt), 599 aa.
Transmembrane assessment:
Count of curated TRANSMEM feature keys in the UniProt feature table
(authoritative line of evidence).
Confirmatory Kyte–Doolittle (1982) hydropathy scan, sliding window 19,
flagging any window whose mean hydropathy reaches the conventional
transmembrane-helix threshold of 1.6.
Nucleotide-binding domains: Walker A / P-loop consensus [AG]-x(4)-G-K-[ST]
and a strict Walker B pattern [ILVFM]{3,4}-D-E.
Iron–sulfur architecture: cysteine inventory and CxxC motifs in the
N-terminal 80 residues, compared against the two UniProt-annotated
4Fe-4S ferredoxin-type domains.

These are heuristics. The Kyte–Doolittle scan and the regular-expression motif
searches are corroborating, not definitive; the curated UniProt feature table
and the experimental literature cited in the review remain the primary evidence.

Results

1. No transmembrane region — ABCE1 is a soluble cytosolic protein

The curated UniProt feature table for P61221 lists zero TRANSMEM features, and an independent Kyte-Doolittle scan (window 19) finds no window reaching the 1.6 transmembrane-helix threshold (global maximum 1.54 at residue 64), confirming ABCE1 has no transmembrane region and is a soluble cytosolic protein.

Metric	Value
UniProt `TRANSMEM` features	0
Max 19-residue Kyte–Doolittle window mean	1.54 (centered on residue 64)
Windows at/above the 1.6 TM-helix threshold	0

The modest hydropathy peak at residue 64 lies inside the N-terminal Fe–S
ferredoxin fold (domain 2, residues 46–75), i.e. it is a buried hydrophobic
core, not a candidate membrane-spanning helix. This is consistent with the
UniProt subcellular location (Cytoplasm) and with the UniProt note that
"The ABC transporter domains seem not to be functional" — ABCE1 is an
ABC-family ATPase that acts on cytosolic ribosomes, not a membrane
transporter. This supports REMOVE of GO:0016020 (membrane) and the
broader interpretation that transporter/membrane-style annotations are
over-annotations driven by the misleading "ABC sub-family E" name.

2. Two nucleotide-binding domains (Walker A / Walker B)

Two Walker A / P-loop motifs were found, at exactly the two ATP-binding sites
annotated by UniProt (BINDING 110..117 and BINDING 379..386):

NBD	Walker A position	Motif	Matches UniProt ATP `BINDING`
NBD1	110–117	`GTNGIGKS`	110–117 ✓
NBD2	379–386	`GENGTGKT`	379–386 ✓

A canonical Walker B (IFMFDE, residues 237–242) is present in NBD1. NBD2's
Walker B (preceding residue ~486) carries an aromatic residue in its
hydrophobic run and so is not captured by the strict [ILVFM]{3,4}DE pattern —
consistent with the well-documented asymmetry/degeneracy of the two ABCE1 NBDs.
Together these confirm two intact nucleotide-binding domains, supporting the
ATP binding (GO:0005524) and ATP hydrolysis activity (GO:0016887)
annotations retained in the review.

3. N-terminal di-cluster [4Fe-4S] (ferredoxin) architecture

The N-terminal region (residues 1-80) contains nine cysteines arranged across the two UniProt-annotated 4Fe-4S ferredoxin-type domains (7-37 and 46-75), consistent with coordination of two [4Fe-4S] clusters rather than mononuclear iron.

Metric	Value
Cysteines in residues 1–80	9 (positions 16, 21, 25, 29, 38, 55, 58, 61, 65)
`CxxC` motif starts	55, 58
UniProt ferredoxin domains	7–37, 46–75
UniProt/InterPro support	SUPFAM SSF54862 "4Fe-4S ferredoxins"; keyword `4Fe-4S`

Eight cysteines are sufficient to coordinate two [4Fe-4S] clusters
(four per cluster); nine are observed. This supports the review's MODIFY of
GO:0005506 (iron ion binding) to the more informative
GO:0051539 (4 iron, 4 sulfur cluster binding).

Limitations

Kyte–Doolittle with a fixed window/threshold is a coarse predictor; a
dedicated tool (DeepTMHMM, Phobius) would be more sensitive. Here it only
corroborates the curated TRANSMEM = 0, which is the authoritative result.
Motif regular expressions can miss divergent sites (as seen for NBD2 Walker B)
and can in principle produce incidental matches; positions were cross-checked
against the UniProt feature table.
Cysteine counting does not prove cluster occupancy; it is consistent with,
not proof of, two [4Fe-4S] clusters. Cluster occupancy is established
experimentally and by structure in the cited literature.