pirl-unc / tsarina / 27284040360

Build # 27284040360

Build Type

push

github

Committed by web-flow

Commit Message

Sanity-check fixes: reference fetch, cutoff semantics, MS classification (#107)

* Sanity-check fixes: reference fetch, cutoff semantics, MS classification

Fixes found in a review pass over the codebase:

- reference_data/cli: `tsarina reference fetch --hpa-version vNN` (fetch-all)
  no longer fails — the HPA version is only forwarded to HPA datasets, not
  the rolling NCBI gene_info (which has no such version). New
  `reference_data.is_hpa_dataset` helper; applied in fetch + path subcommands.
- spanning: MS-tier and predicted-only percentile cutoffs are now inclusive
  (`<=` / `>`), matching the "Maximum percentile" docstrings and the inclusive
  cutoff already used by personalize._assign_tiers.
- cancer_expression: when RNA features are empty but IHC features exist, still
  produce tumor_prevalence_panel_score (default the RNA columns) instead of
  returning an IHC-only frame missing the documented score column.
- personalize: ms-hit merge now validate="m:1" to fail loudly instead of
  silently multiplying candidate rows.
- tiers: cell-line-only MS evidence (peptides with no classifiable source
  flag) is now UNCLASSIFIED_MS instead of being mislabeled NO_MS_DATA.
- peptides: narrow the protein_sequence except blocks to (ValueError,
  KeyError, TypeError) so a real error can't masquerade as a missing
  translation (and silently drop a non-CTA protein from the exclusivity scan).
- selection: drop the dead, misleading `coverage_col` parameter.
- Docstring fixes for restriction return-value sets (tiers, evidence).

Version: 1.10.0 -> 1.11.0

* Add regression tests for PR #107 fixes (UNCLASSIFIED_MS, is_hpa_dataset)

Coverage Stats

2692 of 3632 relevant lines covered (74.12%)

0.74 hits per line

Source File

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

Source Not Available