OncoRef 1.34.0: extracellular sizes, cancer frequency and detailed concerns

The 37-protein membrane table now reports extracellular amino-acid intervals, total external residues and the longest continuous external segment. It links every protein to UniProt, Ensembl and HPA and retains each feature’s evidence code.

UniProt 2026_04 supplies exact external boundaries for 34 of the 37 reference sequences. TEX38, XKR3 and XKR5 have transmembrane annotations without outside-facing boundaries. Their sizes remain unknown.

Thirty-five reference sequences match the selected OncoRef sequence exactly. SPAM1’s selected sequence has the same 455-aa mature chain but a different anchor tail. SLCO1B3–SLCO1B7’s selected 748-aa sequence differs from the 687-aa UniProt reference; its coordinates are not transferred. ZP4’s 487-aa external precursor region is distinguished from its 444-aa annotated mature matrix chain.

Cancer frequency uses HPA 25.1 native pTPM. All 37 proteins have RNA measurements across 21 TCGA cancer types; validation cohorts remain separate. The table shows positive/measured sample counts at ≥1 and ≥5 pTPM, with filters for protein, cohort and cancer. HPA IHC supplies separate staining counts for 14 proteins. Missing measurements remain distinct from measured zero. This panel does not cover every cancer, including NUTM, or measure selected proteoform/surface abundance.

The 29 concern explanations now describe the assays, normal compartments, identity limits and next decisions. FRG2/FRG2B experimental counterevidence remains separate from the five unnamed coding loci. RLN1 nondetection retains the tested sample scope; the reported REL2 serum detection limit is not assigned to REL1. An earlier SRY description is corrected: SRY is the founding SOX-family member, but other SOX-gene studies do not establish SRY expression.

New APIs: cta_literature.surface_domains(), surface_cancer_rna() and surface_cancer_ihc(). Three CSV downloads, native feature evidence, source receipts and a size/prevalence figure accompany the report. The builder runs offline from pinned inputs.

The classification snapshot remains 1.31.0: 602 CTA, 18,255 Unsafe and 1,006 Other. This release adds evidence without changing admission decisions.

Validation checks inclusive coordinates, GPI maturation, sequence mismatches, missing versus zero expression, cohort separation, report filters, offline reproduction and the exported figure.