OncoRef 1.22.0: separate safety and CTA identity

OncoRef 1.22.0 separates normal-tissue safety from CTA identity, with an auditable current selection and reproducible reports.

546 CTA protein-equivalence groups (582 gene loci), 18,031 Unsafe, 20 Neither and 1,266 unassigned across 19,863 human groups. MAGE-A4, NY-ESO-1, PRAME, CTAG2 and XAGE1A/B pass the uniform rules. These are research-policy assignments, not demonstrated clinical safety.

Four independent assessments

1. Safety: absolute RNA limits by tissue tier and usable local IHC, qualifying normal protein and applicable clinical exclusions. Only this assessment can establish Unsafe; Manageable excess blocks CTA without automatically establishing Unsafe.
2. Cancer expression: one measured cancer sample ≥2 TPM or HPA pTPM. No tumor-versus-normal ratio. RNA does not establish protein abundance or presentation.
3. Reproductive/developmental association: reproductive ≥5 nTPM or a supported developmental lineage mean ≥5 TPM, with declared cell-support requirements.
4. Adult restriction: a qualifying association signal exceeds supported adult normal RNA in the same native unit. Vital/Protected/Manageable form the adverse comparator; Reproductive/Disposable do not. CP10K remains safety-only.

All four and required evidence coverage must pass for CTA. Failed admission alone never establishes Unsafe. Neither requires an affirmative measured low-expression or diffuse-Manageable profile; other cases have a null category and an explicit assessment status. Native-unit comparisons remain nominal cross-assay heuristics.

Relative to the preceding 548-group draft, only TGIF2-RAB5IF and ATF7-NPFF move to unassigned on the separate restriction boundary. No safety exclusion changes in that comparison. The previous 1.21.0 release remains frozen with its own smaller selection and sources.

Evidence and provenance

• Uniform source-version selectors across normal references, cancer bundles, source matrices, packaged curated tables and native species assets. Defaults expose registered HPA v25.1 RNA and GTEx v11; the saved selection explicitly pins HPA v23 normal panels.
• Stable/historical identity resolution with conflicting aliases quarantined. Identical-protein members are combined only within compatible measurement contexts.
• Complementary HPA consensus, HPA brain and direct GTEx normal coverage; Petropoulos preimplantation measurements alongside Tyser gastrulation. Optional DICE/cornea gaps remain explicit.
• Every human group has four assessment rows, including passes, in cta_selection.audit().explain(...).
• Native dog nominations remain unassigned/unassessed, and mouse quantitative selection remains unavailable. Human evidence does not transfer through orthology.

Use and migration

from oncoref import cta_selection
selected = cta_selection.results(species="human", category="CTA")
evidence = cta_selection.audit(species="human").explain("PRAME")

Pin oncoref==1.22.0 and policy oncoref.cta-selection.2026-10-29. Historical cta_genes/624-gene compatibility APIs and the v1.20 exact-protein background remain separately versioned; this release does not silently rewrite k-mer backgrounds or product-specific blacklists. Preserve complete group/member identities and unknowns.

Rules · Full report and funnel · Migration notes · PR #643

Assets include the 28-slide PDF and editable deck, a one-page logic PDF, machine-readable decisions/audit, replay reports with provenance receipts, Python distributions and SHA256 checksums. Raw measurement data remain available through the versioned source API; the report archive is not an offline copy of every raw dataset. Figure and slide generators are included in the tagged source and report archive.

Validation: all eight final PR CI jobs passed across Python 3.10–3.14, Windows, minimal installation and lint. The merged tree matches the reviewed head. Strict documentation/package checks, isolated wheel API checks, replay equivalence and rendered-PDF review passed. See release-verification.json and SHA256SUMS.