Published placental and cancer-placenta nominations
For the expanded seven-paper CTA landscape intake and regenerated source plots, see Complete CTA landscape sources. The counts below describe the earlier placental-source addition; the source-specific counts remain valid, while the combined candidate and default totals have since expanded.
Oncoref retains the complete published memberships separately from the HPA CTA candidate table and final default panel. Source membership never overrides the normal-tissue restriction, expression or specificity gates.
| Publication and source | Published | Canonically mapped | Currently coding | Family eligible | HPA pass | Default |
|---|---|---|---|---|---|---|
| Gong 2021, Supplementary Data 5 | 71 | 70 | 69 | 68 | 31 | 26 |
| Gong 2021, Supplementary Data 6 | 74 | 65 | 1 | 1 | 1 | 1 |
| Bradley 2020, Figure 3 | 10 | 10 | 10 | 10 | 1 | 1 |
These source lists overlap and must not be summed. They contain 73 distinct currently protein-coding genes, 31 already in the candidate table and 42 new. The expanded candidate set has 439 rows; the public default has 298 genes, including five new ones: INSL4, GCM1, CYP19A1, HTRA4 and KISS1. This is the September 23, 2026 source-addition snapshot, for oncoref 1.8.202.
In oncoref 1.8.205, TRIM64 was moved to candidate-only status after its nomination and specificity evidence were reviewed. That release's public default had 297 genes; the raw candidate table and the publication-source overlap counts above are unchanged. See the TRIM64 nomination audit and issue #560.
Evidence scope
Gong, The RNA landscape of the human placenta in health and disease, supplies placental RNA enrichment, not tumor-antigen validation. The source annotations are retained: ERVH48-1 was noncoding in S6 and is currently protein-coding; DSCR4 was protein-coding in S5 and is currently lncRNA. TXNRD3NB from S5 and nine S6 entries cannot be mapped by their original Ensembl IDs and remain explicitly unmapped. Historical symbols are not used to guess replacement identities.
The Gong lists support 16/19 prior placental_antigen nominations and 8/9
retained genes. CGB1, CGB2 and CGB7 are not in these lists. The original internal
nomination tags are preserved alongside the new paper-specific tags.
Bradley, Vestigial-like 1 is a shared targetable cancer-placenta antigen expressed by pancreatic and basal-like breast cancers, nominates VGLL1, PLAC1, CGB3, CGB5, IGF2BP3, DEPDC1B, ADAM12, SLC38A9, CAPN6 and MMP11 in Figure 3. HLA-peptide and antigen-specific T-cell validation from this study is recorded for VGLL1 only. Only PLAC1 passes the current default gate; VGLL1 fails the HPA normal-tissue restriction gate and remains available as a sourced candidate.
Data and APIs
get_data("cta-publication-sources"): exact publications, tables/figures, URLs, evidence scope, input hashes and canonical reference hash/release.get_data("cta-publication-membership"): all source rows (including the original 155 placental rows), including unmapped/noncoding entries, source identities and current canonical identities.oncoref.cta_sources.intake_membership()/intake_counts(): auditable mapping, coding, family, HPA and public-default stages.get_data("cancer-testis-antigens"): new coding candidates with source tagsGong2021_placenta_PC,Gong2021_placenta_ncRNAandBradley2020_CPA.
Gong's input hash covers the original XLSX bytes. Bradley's hash covers the transcribed Figure 3 gene-symbol list in the recorded order (UTF-8, one symbol per line with a final newline); it is not a hash of the article PDF.
Reproduce the import
Download the original Gong supplementary workbook using GONG_URL in
scripts/import_cta_publications.py, then run in an environment with oncoref,
openpyxl and the HPA v23 normal RNA/IHC inputs available:
python scripts/import_cta_publications.py --gong-xlsx /path/to/supplement.xlsx
The importer retains every source row, resolves against oncoref's canonical gene space, and adds only mapped, currently protein-coding genes to the HPA table. Existing evidence/filter decisions are unchanged; only new candidates have HPA columns regenerated. Missing RNA stays missing and fails the gate. Unknown canonical transcript IDs, full gene names and functional descriptions for new candidates remain unassigned rather than being inferred. Re-running the import is idempotent.
Pirlygenes renders the source intersections and intake/default funnels from these owner tables, with CSV audits, 300 dpi PNGs and vector PDFs. The local source-specific run records an explicit checkout import; it does not change pirlygenes' released dependency pin or silently rewrite its earlier complete figure batch.
PAGE4 now has a row in the HPA candidate table and is excluded from the default.
cta_candidate_references() therefore omits it from the pending watchlist;
cta_candidate_references(include_in_table=True) retains its historical
literature references and annotations for provenance and clinical/audit joins.
The historical watchlist reports prostate (166.6 nTPM); the canonical CTA
somatic-max scope excludes accessory reproductive tissues and reports smooth
muscle (23.3 nTPM). These are different tissue scopes, not a loss of the prostate
observation.
Closing the CGB provenance gap
The two nomination lists leave CGB1, CGB2 and CGB7 uncovered. Targeted primary papers now give all 19/19 prior placental nominations, including 9/9 retained genes, publication provenance. This statement means expression or nomination support, not that every gene is a validated tumor antigen.
| Source | Genes used here | Evidence and resolution |
|---|---|---|
| Rull and Laan 2005, Methods, Figures 2/4, Table II | CGB1, CGB2, CGB7 | Placental RT-PCR followed by gene-discriminating restriction digestion; CGB2 is distinguishable from CGB1 |
| Rull et al. 2008, Figure 2/Table II | CGB1, CGB2 | Sensitive combined CGB1/CGB2 RNA assay in trophoblastic and normal tissues |
| Kubiczak et al. 2013, Figure 3/Table 1 | CGB1, CGB2 | Combined CGB1-2 RNA signal in ovarian cancers and term placentas; 13/32 cancer positives belongs to the shared assay |
| Bialas et al. 2020, Figures 1-3/Tables 1-2 | CGB1, CGB2 | Separate TCGA gene-level RNA estimates; CGB2 reported in 26 of 33 cancer types; no gene-specific protein-product validation |
| McKellar et al. 2025, Figures 1/2C, Table 1 | CGB7 | Preprint: tumor RNA and CGB7-specific qPCR in SCaBER; immune associations, not HLA/T-cell antigen validation |
cta-gene-publication-evidence records these ten targeted evidence rows;
it is not a transcription of every gene investigated in each article. Each row
retains a source anchor, assay gene scope, review status, limitations and NCBI
BioC input checksum. The raw full texts are retrieved through NCBI's public
BioC endpoint; they are not redistributed in the package.
oncoref.cta_sources.gene_publication_evidence() exposes these rows;
placental_source_coverage() reconciles the historical list to publications and
the default set. The importer attaches their tags without changing HPA values.
CGB2 stays in the default; CGB1/CGB7 stay excluded. No CGB-specific protein,
HLA-peptide or T-cell validation is asserted from this evidence. In particular,
combined assays and beta-hCG family protein measurements cannot establish a
unique CGB2 protein or epitope.