Skip to content

Load short expression CSV and TSV tables consistently - #487

Merged
iskandr merged 1 commit into
masterfrom
fix/486-short-expression-tables
Oct 7, 2026
Merged

iskandr merged 1 commit into
masterfrom
fix/486-short-expression-tables

Conversation

@iskandr

@iskandr iskandr commented Oct 7, 2026 •

Copy link
Copy Markdown
Contributor

Valid expression CSV tables with a header and one patient measurement were parsed as TSV because delimiter inference required three lines. Infer from the available informative lines instead; skip comments and blank lines, and retain an explicit error for files without any header/data lines. Header-only tables work with declared value columns; inferred numeric columns still require measurements.

Both file and CLI-spec loaders run the same CSV/TSV battery with zero, one and multiple rows. A composed regression loads one expression feature, joins it to predictions, changes selection with abundance, and saves/replays the selection decision. Bumps Topiary 5.94.2 to 5.94.3.

Fixes #486. Unblocks openvax/vaxrank#585 without a consumer-specific delimiter workaround.

Validation: ./lint.sh passed; 17 focused regressions passed; the complete ./test.sh passed 5,469 tests with 20 optional skips in 709.91 seconds. All exact-head GitHub CI checks passed, including Python/pandas matrices and Isovar/PirlyGenes integration. Local HTTP retry tests required localhost networking outside the sandbox.

The full gate also exposed high-depth RNA support-processing cost in an existing mitochondrial fixture; recorded separately at openvax/isovar#463. No Isovar behavior, read limits or consumer workaround changed in this PR.

@coveralls

Copy link
Copy Markdown

Coverage Status

coverage: 93.527% (+0.2%) from 93.289% — fix/486-short-expression-tables into master

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.

Expression loader misreads CSV tables with fewer than two data rows as TSV

2 participants