OncoRef 1.35.0: MANE-first reference proteins

Reference selection now uses MANE Select first, Ensembl canonical second. The selected CDS must be complete on the primary locus. Incomplete annotation, conflicting designated sequences and unresolved coding evidence retain a candidate for review without an approved default. Length and alternate-locus sequences do not override this choice.

The catalog contains 607 CTA, 18,259 Unsafe and 1,027 Other proteins across 19,893 protein identities. Eight proteins newly qualify under the representative rule; GDPD4, SLCO1B3–SLCO1B7, FRG2 and FRG2B move to translation review. CDY1 and CDY1B now have separate defaults. RBMY1F and RBMY1J retain their shared identity.

The method report explains incomplete cases and how additional evidence can support, hold, resolve or prefer an exact reference sequence. Its searchable table covers all 20,099 classified loci, including 563 awaiting representative review. Downloads retain transcript versions, support levels, alternatives, hashes and every change from the previous catalog. MANE and coding annotation do not establish tumor-specific protein production.

Normal and tumor expression are reassessed after membership changes. Protein summaries are rebuilt for 142 cancer cohorts; gene-level results reproduce exactly. NUTM's stored input contains negative or infinite values in a declared TPM field. Its historical protein summaries are marked unavailable and rejected by the reader. Source repair remains in #604 and #665.

The sequence bundle retains all 245,535 reference peptide records and their source occurrences, with updated default roles and review requirements. The membrane-coordinate audit transfers coordinates only to an exact approved sequence. Earlier 602-protein literature reports remain dated snapshots.

New APIs: transcript_selection.records() and transcript_selection.evidence(). cta_selection.profiles() distinguishes sequence qualification, representative approval, candidate sequences, selected transcripts and reviewed findings.