CTA definitions — 25 October 2026
213 CTA genes, 2179 Neither and 17705 Unsafe across 20,097 canonical human genes. Among 2,540 mapped source nominations: 213 CTA / 741 Neither / 1586 Unsafe.
Plain-language rules and complete threshold table · All candidates · All genes · CTA · Neither · Unsafe
The updated policy applies reviewed literature as biological findings, not gene-name overrides. The unchanged measurement facts plus the previously accepted findings reproduce all 20,097 old categories exactly. New reviewed findings change 186 assignments from the released 399-gene CTA set (6 to Unsafe, 180 to Neither). RNA thresholds and tissue mappings are unchanged.
Named targets — retained and excluded
| Gene | Category | Detail |
|---|---|---|
| MAGEC2 | Neither | Conflict |
| CTCFL | Neither | Conflict |
| MAGEA10 | Neither | Conflict |
| AKAP4 | Neither | Conflict |
| MAGEA4 | Neither | Conflict |
| MAGEC1 | Neither | Conflict |
| SSX5 | CTA | Supported |
| CT55 | CTA | Supported |
| SAGE1 | Neither | Conflict |
| CTAG1B | CTA | Pass |
| PRAME | Unsafe | Normal |
| MAGEA1 | Neither | Conflict |
| CT83 | CTA | Pass |
| XAGE1A | CTA | Pass |
| XAGE1B | CTA | Pass |
| GAGE1 | CTA | Supported |
| SPANXB1 | CTA | Pass |
| SSX2 | CTA | Pass |
| SSX4 | CTA | Supported |
| CTAG1A | CTA | Pass |
| DAZL | Neither | Cellular |
| TDRD1 | Neither | Cellular |
| CTAG2 | Neither | Variable |
| BRDT | Neither | Cellular |
| HORMAD1 | Neither | Cellular |
| FOXR2 | Neither | Antibody |
| CTAGE1 | Neither | Antibody |
| ABCB5 | Unsafe | Normal |
| SPAG9 | Unsafe | Protein |
| DEPDC1 | Unsafe | Protein |
| SPA17 | Unsafe | Protein |
| ROPN1 | Unsafe | High RNA |
| TRIP13 | Unsafe | High RNA |
| BIRC5 | Unsafe | Protein |
| ACRBP | Unsafe | High RNA |
| TTK | Unsafe | Protein |
| KIF20A | Unsafe | Protein |
| IL5 | Unsafe | Normal |
| ROPN1B | Unsafe | High RNA |
| TEX101 | Unsafe | High RNA |
| IGF2BP3 | Unsafe | Protein |
| KIF20B | Unsafe | Protein |
| NUF2 | Unsafe | High RNA |
| LY6K | Unsafe | High RNA |
| PTTG1 | Unsafe | Protein |
| IL13 | Unsafe | Normal |
| PAPPA | Unsafe | Normal |
| CLDN6 | Unsafe | High RNA |
| MAGEA12 | Unsafe | Safety |
| MAGEA3 | Unsafe | High RNA |
| GAGE12F | Unsafe | High RNA |
| PAGE2B | Unsafe | High RNA |
Every changed gene has a review record with source, species, reference, assay, identity/specificity judgement, review depth, limitations and linked issue. Abstract-level decisions are marked. Unsafe means exclusion from unrestricted gene-level CTA targeting, not demonstrated clinical injury. Neither conflicts have a completed curation decision; biological uncertainty remains. They need evidence that resolves the conflict before reconsideration.
Evidence and follow-up
Decision facts · Reviewed literature · Changed rows · Follow-up questions · Named CTA checks · Literature search and review coverage
CT55 remains CTA through affirmative reproductive and tumor-recognition evidence below the RNA floor; normal gates still apply. CTAG1A/B and XAGE1A/B remain CTA. CTAG2 remains Neither / Variable and FOXR2 Neither / Antibody. MAGEA4 is now Neither / Conflict; PRAME is Unsafe because an independent human adrenal-protein observation is accepted. This does not relabel their selected therapeutic products as clinically unsafe.
All 178 completed antibody adjudications, evidence and reconsideration requirements. PLCZ1's pancreatic paper measures its RNA, while its immunostaining detects CFTR; the erroneous protein interpretation is rejected and the qualitative RNA finding is held separately.
The literature audit is incomplete: 32 genes received new primary-paper assessment, with 46 specific notes including prior reviews. All 178 raw-antibody discrepancies remaining in the local 391-gene CTA draft received documented adjudications: 177 Neither and 1 Unsafe. The two other genes in the original 180-gene queue were already addressed by the eight-gene review. 0 unresolved raw-antibody cases remain retained. Review completion does not mean the molecular identity of every stain was resolved. The retained CTA set remains a research selection, not validated clinical safety. No new continuous breadth gate or calibrated RNA uncertainty model is claimed.
No remaining CTA or Neither has an audited non-permitted/unsited TPM or nTPM observation >30. Missing observations remain gaps; this is not proof of absence in unmeasured cells. Original source versions, native units and donor limitations are preserved in the RNA audit.
Exact proteins
Complete exact-sequence groups: 359 CTA / 3,230 Neither / 126,396 Unsafe. Priority for a protein encoded by multiple genes is Unsafe → Neither → CTA. The additional 1,721 reference loci retain their measured decisions, reissued under this policy. Neither is excluded from the default forbidden union.
Sequence recipe · Reference · Release asset inventory and SHA-256 · Additional loci
Use the matching v1.20.0 release assets and their recorded checksums. Earlier FASTAs retain their dated policies and must not be combined with the new gene list. The historical 624-gene CTA API is unchanged; downstream applications must explicitly choose the current target_sets policy.
Reproduction
.venv/bin/python scripts/build_cta_policy.py --output /tmp/cta-policy-report --assets /tmp/cta-policy-proteins --sequence-input /path/to/pinned-original-sequence-assets
Use empty output directories. Add --publish-dataset only to update the repository's packaged current data after all invariants pass. The builder verifies pinned input and original sequence hashes, replays the baseline, evaluates every reason, and checks the >30 invariant before publishing.
Policy · Counts · Input/code/output hashes. Historical reports remain frozen.