CTA recalibration and complete intake — 21 October 2026

Policy oncoref.cta-ihc-adaptive-screen.2026-10-21.1. 13 publication sources, 3,906 distinct intake identities and 2,538 mapped protein-coding candidates. Human HPA v23 measurements remain pinned.

Current rules

  • Reproductive admission requires valid RNA >0 nTPM OR positive reproductive IHC. A measured zero does not qualify; unavailable observations remain missing.
  • Disposable expression is exempt from exclusion and cannot supply the reproductive reference or admission evidence.
  • Selectivity = Rmax / (Rmax + tissue RNA), with Rmax the highest valid raw reproductive nTPM for the same gene/source/version/species.
  • Each Manageable, Protected and Vital tissue must pass both its absolute ceiling and selectivity minimum. Any positive IHC, including Low, still excludes.
  • The same complete routine local negative IHC selects both thresholds, using the weakest reliability. Missing, uncertain, incomplete or unusable IHC receives baseline limits. Zero/missing reproductive RNA leaves selectivity unresolved.

Maximum nTPM / minimum selectivity

IHC Vital Protected Manageable
Missing 1 nTPM / 75% 3 nTPM / 70% 5 nTPM / 65%
Uncertain 1 nTPM / 75% 3 nTPM / 70% 5 nTPM / 65%
Approved 2 nTPM / 70% 4 nTPM / 65% 10 nTPM / 60%
Supported 3 nTPM / 65% 6 nTPM / 60% 15 nTPM / 55%
Enhanced 6 nTPM / 60% 12 nTPM / 55% 30 nTPM / 50%

The user approved this target-informed research recalibration. The previous 99% minimum required a 99-fold reproductive/tissue ratio and could exclude 0.1 nTPM observations despite a 1 nTPM absolute ceiling. The new percentages were selected for the pairwise metric while retaining absolute and IHC gates; they are not validated safety limits. Protected's baseline changes generally from 2 to 3 nTPM, which admits CT83's salivary 2.6 nTPM. There are no gene-specific numerical exemptions. Named-target recovery is a calibration check, not independent validation.

Policy comparison

scenario candidates evidence_pass retained without_ihc partial_ihc complete_routine_ihc
previous_released 2537 430 377 176 197 4
admission_only 2538 490 409 208 197 4
selectivity_revised 2538 757 642 369 269 4
grid_revised 2538 762 647 374 269 4
current_adaptive 2538 762 649 375 270 4
current_fixed 2538 719 606 375 228 3

The first row is the archived v1.11.0 cohort. Subsequent rows include the new SPAG9 candidate. admission_only changes reproductive admission; selectivity_revised then changes the percentage grid; grid_revised also changes Protected's baseline; current_adaptive additionally applies explicit holdback reviews. current_fixed uses baseline RNA/selectivity values without IHC relaxation.

Policy comparison

Current funnel

stage remaining
Published source identities 3906
Mapped protein-coding candidates 2538
Positive reproductive evidence 2471
Usable resolved evidence and routine RNA coverage 2071
No positive IHC outside permitted tissues 1600
Absolute RNA ceilings 851
Pairwise selectivity 762
Family eligibility 759
Nomination and current review eligibility 649

This sequential ordering attributes each removal to its first failing stage. Failure reasons overlap; every reason remains in the candidate decisions. An intake identity lacking a canonical mapping is visible upstream and is unassessed, never assigned zero RNA.

Source-to-screen funnel

Requested targets and review provenance

Symbol retained failure_reasons screen_review_applied
CT83 True False
CTAG2 True True
CTCFL True False
GAGE1 True False
MAGEA10 True False
MAGEA11 True True
MAGEA12 True False
MAGEA6 True False
SAGE1 True False
SPAG9 False reproductive_selectivity_below_threshold;rna_above_tissue_ceiling;positive_ihc_outside_permitted False
SSX4 True False
SSX5 True False

CTAG2's old categorical RNA holdback is explicitly superseded for this screen, conditional on all current expression and coverage gates. Its pinned heart 5.1 nTPM and Enhanced Not detected cardiomyocyte staining remain visible. The heart evidence review retains unresolved cell origin, antigen-presentation questions and confounded clinical cardiac-event reports. MAGEA11 receives the same review of its historical trace-RNA holdback. Neither adjudication bypasses positive IHC or a numerical failure. The frozen core/extended panels retain their original reviews.

BAGE, GAGE2B and SPAG9 intake

Symbol intake_identity screened screen_status
SPAG9 ENSG00000008294 True excluded
BAGE NCBIGene:574 False unassessed_no_canonical_mapping
GAGE2B NCBIGene:645037 False unassessed_no_canonical_mapping

BAGE and GAGE2B have official human NCBI identities, confirmed in a checksum-pinned snapshot, but lack a usable mapping in the pinned Ensembl 115 canonical space. Their memberships, citations and identity reviews are preserved in the funnel. No paralog is substituted, and their HPA normal-tissue expression is unassessed here. Jamin's GAGE2B nomination uses shared microarray probes; it does not establish locus-specific expression. SPAG9 enters the coding screen from its primary ovarian-cancer study and receives the same HPA gates as every other gene.

New primary sources: Boël et al. 1995, BAGE and Garg et al. 2007, SPAG9. Identity authorities: NCBI BAGE, NCBI GAGE2B. Source assays and limitations are preserved separately from identity resolution.

Coverage

Complete routine RNA is required. IHC-unavailable and partial-IHC candidates remain explicitly distinguished from routine-complete observations; none proves exhaustive protein or peptide-HLA absence. The routine panel contains 50 RNA and 80 IHC tissue/cell contexts. Missing developmental and anatomical compartments remain unassessed. Human assay limits do not transfer to dog or mouse. See category definitions and mappings.

Downloads

Reproduce with python scripts/cta_recalibration_report.py --out <empty-directory> at this implementation revision. Existing candidate measurements are reused from the verified archive; the added SPAG9 observations come from the same checksum-pinned HPA v23 inputs.