Skip to content

Readable CTA curation, 2 nTPM floor and disjoint gene/protein sets - #614

Merged
iskandr merged 1 commit into
mainfrom
normal-sequence-sets
Oct 21, 2026
Merged

iskandr merged 1 commit into
mainfrom
normal-sequence-sets

Conversation

@iskandr

@iskandr iskandr commented Oct 21, 2026

Copy link
Copy Markdown
Contributor

Summary

  • Double the approved absolute RNA ceilings: Vital 2/4/6/12, Protected 6/8/12/24, Manageable 10/20/30/60 nTPM. Keep reproductive admission >0, the approved selectivity percentages, coverage requirements and positive-IHC veto.
  • Recompute all 20,097 human coding genes and the complete CTA intake. The current curated set retains 746 genes (98 gained, none lost); XAGE1A and XAGE1B both pass. Fixed/adaptive/zero-measured-outside-RNA comparisons retain 708/746/216.
  • Replace report jargon with Clinical evidence / Result / Reason, add a searchable gene explorer, current publication counts, clinical evidence links, tissue measurements, comparison downloads, a funnel figure and navigation.
  • Clarify GAGE2B's archived Ensembl ID versus its ambiguous current GAGE2A cross-reference. Preserve BAGE/GAGE2B as unassessed rather than borrowing paralog expression.
  • Add disjoint gene and complete exact-protein partitions using all Ensembl 115 isoforms, full occurrence provenance and Normal > Unassessed > Reproductive > Other precedence. XAGE1A/B share 69- and 81-aa complete proteins in Reproductive. Partial translations keep their gene group; only Normal/Unassessed records enter additional background.
  • Publish small report metadata in docs and place the 134 MiB of sequence tables/FASTA exports in v1.13.0 release assets. Historical reports remain immutable. Heavy DATA_VERSION remains 5.23.26.

Verification

  • 117 policy/partition tests and 188 identity/clinical/source/manifest tests passed.
  • Ruff lint/format and strict MkDocs build passed.
  • Rechecked all 2,540 candidates' adaptive RNA/selectivity flags against raw pinned HPA v23; every decision matches the genome-wide report.
  • Verified every new report input/output checksum, sequence disjointness, exact XAGE1A/B protein sharing, and partial-background routing.
  • Executed the explorer JavaScript with a minimal DOM harness: initial render, pagination, clear, gene/alias search, inclusion/clinical filters and no-results behavior pass. All local report links resolve. Funnel figure visually inspected.
  • Browser preview could not run: no browser provider was available and the native computer-use service failed to start. Responsive layout is implemented but has not received live browser visual QA.

Interpretation

These are user-approved research parameters, not validated clinical safety limits. Numerical passes do not erase measured trace expression. Positive normal IHC and the external MAGEA12 safety hold remain exclusions. Full-protein sets are disjoint; k-mers still require actual substring subtraction. Protein groups inherit gene-level expression and do not establish isoform-specific expression.

@iskandr
iskandr merged commit d9b94db into main Oct 21, 2026
8 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant