citation	evidence_type	direction	claim_tested	key_finding	context	confidence_limitations
UniProt D3VIU4 (database)	computational/database	refutes	D3VIU4 is a channel	Submission name "Cysteine transport protein (ABC superfamily, peri_bind)"; gene fliY; only GO = C:periplasmic space (GO:0030288); no channel/transport MF	X. nematophila proteome, inferred from homology	High; database-level, but consistent with all domain evidence
Pfam PF00497 (SBP_bac_3) / InterPro IPR001638	structural/evolutionary	refutes	Channel/TM architecture	Single solute-binding protein family 3 domain (res 40-259); this is a soluble periplasmic ligand-binding fold, not a membrane channel	Domain assignment	High; family is defined as extracytoplasmic solute receptors
CDD cd13711 (PBP2_Ngo0372_TcyA)	structural/evolutionary	refutes	Channel identity	Matches TcyA cystine-binding periplasmic protein subfamily	Conserved-domain database	High
SUPFAM SSF53850 / Gene3D 3.40.190.10	structural/evolutionary	qualifies	Fold relationship	"Periplasmic binding protein-like II" fold = the extracellular ligand-binding module also present in ionotropic glutamate receptors; explains ProtNLM2 confusion but D3VIU4 lacks any fused ion-channel/TM domain	Structural classification	High; fold homology to iGluR LBD is the likely misassignment source
Computed KD hydropathy (this study, provenance)	computational	refutes	TM pore-forming architecture	0 transmembrane segments (window 19, thr 1.6); only hydrophobic peak res 11-18 lies inside signal peptide 1-28; 0 cysteines	Sequence-based, 262 aa	High for absence of TM; window method, not TMHMM/DeepTMHMM
UniProt features	localization	refutes	Membrane-embedded channel	Signal peptide 1-28 (cleaved), mature chain 29-262 = secreted/periplasmic soluble protein	UniProt annotation	High
PMID:25837721	direct assay / mutant phenotype	refutes	D3VIU4-type protein is a channel	FliY is "the periplasmic L-cystine-binding protein"; membrane permease is YecS; FliY-YecSC ABC importer imports L-cystine (Km=110 nM)	E. coli functional genetics	High; ortholog (E. coli FliY), not X. nematophila directly
Orthology to E. coli FliY P0AEM9 (this study + PMID:25837721)	structural/evolutionary + direct assay	refutes	Channel vs binding-protein identity	Needleman-Wunsch global alignment: 131/240 = 54.6% identity full-length to characterized periplasmic L-cystine-binding protein FliY; transfers soluble periplasmic cystine-binding function	Sequence comparison to E. coli functional data	High; ortholog-level function transfer at >50% identity
AlphaFold AF-D3VIU4-F1 v6 (this study)	computational/structural	refutes	TM pore-forming architecture	Mature core 29-262 = compact globular domain (Rg 17.8A, aniso 1.17, pLDDT 96.0); signal peptide 1-28 low pLDDT 46.5 protruding 58.8A; most hydrophobic 19-mer meanKD 1.58 (<1.6) => no TM helix	AlphaFold DB predicted structure	High confidence model (pLDDT>90); prediction not experimental structure
ProtNLM2 prediction (seed)	computational	competing	Ligand-gated ion channel (GO:0015276)	Automated name/annotation model prediction under evaluation	ML text model	Low; refuted by all lines above