Forty GOA rows were reviewed using gene-specific primary evidence. PMID:10859168(https://pubmed.ncbi.nlm.nih.gov/10859168/) establishes the unfused-myoblast phenotype and membrane localization; PMID:19270174(https://pubmed.ncbi.nlm.nih.gov/19270174/) identifies Sns as the ligand of Kirre/Roughest; PMID:18971929(https://pubmed.ncbi.nlm.nih.gov/18971929/) and PMID:19515699(https://pubmed.ncbi.nlm.nih.gov/19515699/) establish the nephrocyte diaphragm function. PMID:24485456(https://pubmed.ncbi.nlm.nih.gov/24485456/) directly measures Drosophila adhesion heterocomplexes despite its worm-focused title. Binding to Mec2 and Unc-5 is kept noncore; Roughest binding is specified as cell adhesion molecule binding. Abstract-only evidence leaves homophilic adhesion and the specific actin-focus result unresolved, without rejecting the experiments.
The sequence analysis finds that selected Q0E9F2 isoform B contains all 1479 residues of RefSeqA identically, plus 63 residues around the membrane-entry region. The insert has its own predicted transmembrane segment. This is a concrete topology concern, not absence of a separate assay for every isoform. Gene-level annotations are reviewed as gene-level evidence; the two broad developmental ProtNLM claims remain UNC for this exact longer sequence pending topology/functional assessment. Membrane localization remains supported and is LSP. These findings do not imply wrong predictor input or prove the longer isoform inactive.
The selected frozen UniProt record is retained in sns-uniprot-source.json, with the complete original prediction metadata in sns-predictions-source.json. It maps the exact accession to this FlyBase gene; current sequence identity does not establish the historical predictor input. Gene-level experiments support conserved functions unless a relevant isoform difference is established. Falcon research was requested with perplexity-lite fallback alongside publication caching. Primary publications and sequence observations, rather than AI summaries or ARBA assertions, support the curated conclusions.
The wrapper timed out and its fallback reported insufficient quota, but the original Falcon job later produced a complete report. The report and primary sources were inspected; no duplicate provider submission was made. The report correctly summarizes gene-level adhesion/fusion and nephrocyte functions, and cautions that foundational constructs are not isoform-B-specific. The exact hydrophobic insertion remains established by the local analysis; report statements about the standard single-pass model do not resolve its topology.