pirl-unc / tsarina / 25673438197

Build # 25673438197

Build Type

push

github

Committed by web-flow

Commit Message

Surface TCGA pan-cancer RNA prevalence in evidence and panel summary (#67)

The bundled HPA cancer RNA-seq prevalence table already includes
cohort=TCGA rows for 21 TCGA project equivalents (BRCA, LUAD, ...),
but tsarina previously consumed those values only inside the
tumor_prevalence_panel_score ranker. Callers using CTA_detailed_evidence()
saw no per-gene TCGA prevalence, and the per-CTA panel summary table
gave no signal of which CTAs were broadly expressed in tumors versus
which were testis-restricted with low tumor expression.

Adds:
- cta_tcga_expression_features() in tsarina.cancer_expression — one row
  per CTA with TCGA-only sample counts, pan-cancer prevalence at
  TPM>=1 and TPM>=5, and the TCGA cancer type that maximizes prevalence
  at the lower threshold.
- CTA_detailed_evidence() now attaches the nine TCGA columns by joining
  on the stripped Ensembl gene id, with zero/empty defaults for genes
  without TCGA measurements.
- panel_summary() accepts an optional tcga_features arg and adds
  tcga_pan_prevalence_tpm_ge_1 / tcga_pan_prevalence_tpm_ge_5 /
  tcga_max_ptpm / tcga_top_cancer_type to each per-CTA row, taking the
  max across slash-joined CTA members.
- format_panel_summary() renders three new columns ("TCGA >1 TPM",
  "TCGA >5 TPM", "TCGA top type") with cancer types abbreviated to
  TCGA project codes (SKCM, LUAD, ...) to keep the table narrow.
- spanning_pmhc_set() best-effort loads the bundled TCGA features so
  the CLI panel output picks up the new columns automatically.

Also modernizes pyproject license metadata (#49): switches
project.license from a TOML table to the SPDX string "Apache-2.0",
adds project.license-files = ["LICENSE"], and drops the deprecated
"License :: OSI Approved :: Apache Software License" classifier so
setuptools no longer emits a deprecation warning during deploy.

Closes #55.

v1.3.0

Coverage Stats

2383 of 3386 relevant lines covered (70.38%)

0.7 hits per line

Source File

Press 'n' to go to next uncovered line, 'b' for previous

Source Not Available