Fix crash from factor sample.class and sparse tmb tables (#11)

[Arvindiyer]

Summary

• get.blocks() lost sample names when sample.class/alteration.class was a factor, because which() drops names on a named-factor comparison but not a named-character one. The resulting unnamed blocks propagated into 0-column matrices in template.obj.gen()/generateW_block(), which crashed retrieveOutliers() with "replacement has length zero" several steps downstream.
• Separately, new.AL.general() summed am$tmb tables by raw vector position, so a tmb table covering only a subset of samples (e.g. only samples with >=1 mutation of a given type) got silently recycled instead of matched by sample identity, corrupting TMB totals without erroring.
• Both are now handled at ingestion in new.AL.general(): factor covariates are auto-coerced to character, and every tmb table is aligned to the full sample set by name with missing samples zero-filled. Unknown sample ids or duplicate entries in a tmb table now raise a clear error instead of being silently mishandled.

Test plan

• devtools::test() — 24/24 passing, including the existing golden LUAD regression test unchanged
• devtools::check() — 0 errors, 0 warnings, 0 notes
• New tests added in tests/testthat/test-new.AL.general-robustness.R covering factor sample.class, sparse tmb zero-filling, and the new unknown-sample-id error
• Re-ran the reporter's original repro shape (factor sample.class + sparse tmb) — completes cleanly with no error/warning

Fixes #11

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 4c00fa1e-b78d-474f-8b3e-3c48cf4d5d34

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}