OncoRef 1.36.0: cancer-only audit and uniform decisions

The individual cancer-only audit examines 47 proteins with strong cancer RNA and zero recorded adult reproductive bulk RNA. It starts from the 66 matches to the existing normal-low tumor-expression screen; the other 19 have measurable reproductive bulk RNA. No member establishes cancer-only protein production. All 47 have some developmental RNA signal, and reproductive-cell coverage remains incomplete.

The reviewed calls are 3 Unsafe, 36 Other:Review and 8 Other:Expression. The report links every protein to normal-tissue observations, olfactory assay results, individual supporting tumor observations, Ensembl, HPA and primary papers. It includes the 2026 fetal olfactory atlas and keeps human, animal, endogenous and recombinant experiments distinct. HPA's uncertain shared antibodies remain review evidence. Missing assays retain missing values.

Two assay details change interpretation: the olfactory study's calibrated RNA-copy cutoff differs from raw Cq detection, and two probes for one gene do not provide independent donors. Cross-reactive receptor pairs are retained explicitly. KRTAP21-1 has HPA MS protein-existence evidence without tissue/peptide attribution in the XML; it is preserved separately from the retrieved UniProt annotation.

cancer_only.records() exposes all 47 individual reviews, while cancer_only.audit() returns their definitions, counts and source links. cta_selection.decisions() uses CTA, Other and Unsafe with three Other reasons: Review, Expression and Sequence. Every detailed reason and member gene is retained. The original release labels remain available through results().

This package release publishes code and reviewed evidence. The classification snapshot and sequence bundle remain 1.35.0, with 607 CTA, 18,259 Unsafe and 1,027 Other; expression data remain 5.23.27. The explicit curation replay has 600 CTA, 18,259 Unsafe and 1,034 Other, including the seven previously identified normal-expression holds. New cancer-only-audit findings refine Other review reasons and introduce no additional CTA removal or Unsafe assignment. The next scientific data rebuild must publish updated classifications and sequence roles together.