Curation summary
Policy oncoref.target-selection.2026-11-10.1, 19,893 human protein records. 626 CTA, 245 Cancer Antigen only, 31 Reproductive Antigen only. These are outputs of screening rules, not validated tumor antigens. Every figure is drawn from the rebuilt catalog; thresholds are read from the policy constants.
Rule-derived sets and published nominations
Of 626 CTA, 402 are published nominations and 224 are not. 1,801 published nominations are in neither set: 1,758 of all 2,233 nominations are Unsafe, 36 Other:Review and 7 Other:Sequence. A published nomination is an annotation; it never overrides an exclusion.
Funnel by filtering idea
A protein leaves at the first idea that removes it, so later steps show only what each idea adds. The order is chosen so that every step removes something; the layers overlap heavily, as the next section shows.
| Filtering idea | Removed here | Remaining |
|---|---|---|
| All assessed proteins | 0 | 19,893 |
| Bulk tissue RNA above the tissue ceiling | 17,522 | 2,371 |
| Cell-type cluster RNA (HPA single cell) | 424 | 1,947 |
| Sorted immune-cell RNA | 96 | 1,851 |
| Rare-cell RNA established by donor | 107 | 1,744 |
| Normal-tissue protein (IHC) | 50 | 1,694 |
| Curated or published normal expression | 37 | 1,657 |
| Adult cornea RNA | 5 | 1,652 |
| Incomplete coverage, identity or translation | 693 | 959 |
| Nested or Unsafe-contained sequence | 42 | 917 |
| No qualifying expression or bounds fail | 15 | 902 |
| Filtering idea | Removed here | Remaining |
|---|---|---|
| Published CTA / oncofetal nominations | 0 | 2,233 |
| Bulk tissue RNA above the tissue ceiling | 1,464 | 769 |
| Cell-type cluster RNA (HPA single cell) | 174 | 595 |
| Sorted immune-cell RNA | 25 | 570 |
| Rare-cell RNA established by donor | 68 | 502 |
| Normal-tissue protein (IHC) | 19 | 483 |
| Curated or published normal expression | 7 | 476 |
| Adult cornea RNA | 1 | 475 |
| Incomplete coverage, identity or translation | 36 | 439 |
| Nested or Unsafe-contained sequence | 7 | 432 |
Steps that remove no published nomination are omitted from the second funnel. All proteins · Published
Normal-expression layers
| Layer | Unsafe proteins flagged | Only reason |
|---|---|---|
| Normal bulk | 17,522 | 560 |
| HPA tissue clusters | 17,053 | 268 |
| Adult cornea | 12,095 | 5 |
| DICE immune | 11,632 | 45 |
| Monaco immune | 11,264 | 19 |
| HPA normal protein | 10,157 | 47 |
| HPA immune | 7,973 | 14 |
| Established rare normal-cell RNA | 129 | 106 |
| Reviewed normal-cell knowledge | 32 | 31 |
| Accepted normal-protein evidence | 8 | 2 |
| Published normal-tissue expression | 7 | 6 |
Thresholds
Ceilings are tier base × 2 per negative-IHC reliability step (Vital 4, Protected 8, Manageable 16 without IHC credit). Exceeding a Vital or Protected ceiling is Unsafe; Manageable is Unsafe only above 128. These are screening conventions, not measured injury thresholds.
The second table counts recorded bulk-RNA peak observations, not proteins: one protein can contribute a peak per source.
Expression distributions and where the thresholds fall
One value per protein on each source's own scale; scales are never pooled. Counts use a compressed axis so the small candidate group stays visible.
Cancer rule
Rare-cell RNA
Inside the later stages
- Required normal coverage incomplete: 400
- Identity or translation unresolved: 102
- Normal signal blocks admission (not Unsafe): 159
- Patient identity unresolved in tumor samples: 32
A protein can use several routes, so route counts exceed set sizes.