Remove mix-max dead code and --method CLI flag

[wsnoble]

Summary

• Removes the MixmaxConfidence class (~177 lines) from confidence.py — the mix-max method was previously removed but its class was left behind
• Removes mixmax(), estimate_pi0(), and calculate_mixmax_qval() from qvalues.py (~211 lines) — these only served mix-max
• Removes the --method/-m CLI flag from params.py — its only remaining choice was "tdc", making it a no-op
• Removes the method parameter from PsmDataset.assign_confidence() and dataset.py's internal dispatch dict, replacing method-based dispatch with a direct TdcConfidence(...) call (the backend parameter added in PR29 is preserved)
• Cleans up unused imports: itertools in utils.py, ClassVar in dataset.py, re in three parsers, read_txt/read_pepxml in msfragger.py
• Fixes test fixtures to use Path(__file__).parent.parent / "data" so tests pass regardless of the working directory from which pytest is run

Test plan

• Full test suite passes: pytest tests/ → 88 passed, 2 skipped
• All method="tdc" kwargs removed from all test files (test_confidence.py, test_edge_cases.py, test_writers.py)
• No mixmax fixtures or tests remain in test_qvalues.py or test_confidence.py
• numba import retained in qvalues.py — still used by _fdr2qvalue

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Team

Run ID: 1c2a2fdd-9978-4ed0-8e73-45ad09bd46f7

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch remove-dead-code

---

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.}

[Copilot]

Pull request overview

This PR removes the previously-deprecated MixMax confidence/q-value implementation and simplifies the confidence estimation API/CLI to the single remaining method (TDC), while updating tests and fixtures accordingly.

Changes:

• Deleted MixMax-related code paths (MixmaxConfidence and mixmax q-value helpers) and removed method-based dispatch in PsmDataset.assign_confidence().
• Removed the no-op --method/-m CLI flag and updated CLI dispatch to match the new API.
• Updated unit tests/fixtures to remove method="tdc" usage and to use a stable DATA_DIR path independent of the current working directory.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tests/unit_tests/test_writers.py	Removes method="tdc" from confidence assignment calls in writer tests.
tests/unit_tests/test_qvalues.py	Removes MixMax imports and MixMax test coverage; keeps TDC tests.
tests/unit_tests/test_parquet.py	Removes method="tdc" from Parquet writer tests.
tests/unit_tests/test_edge_cases.py	Removes method="tdc" from edge-case confidence assignment tests.
tests/unit_tests/test_duckdb.py	Updates backend validation test to check unknown backend rather than invalid method.
tests/unit_tests/test_confidence.py	Removes MixMax confidence tests and method="tdc" usage; keeps TDC tests.
tests/conftest.py	Introduces DATA_DIR fixture root to make test data paths robust to pytest CWD.
crema/utils.py	Ensures parse_psms_txt doesn’t use the pyarrow engine when skiprows is required.
crema/qvalues.py	Removes MixMax q-value estimation functions, leaving TDC and helper _fdr2qvalue.
crema/params.py	Removes the --method/-m CLI argument.
crema/dataset.py	Removes method dispatch and always instantiates TdcConfidence for pandas backend; preserves backend selection.
crema/crema.py	Stops passing args.method into assign_confidence now that the flag/parameter is removed.
crema/confidence.py	Removes the MixmaxConfidence class implementation.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[bmx8177]

@wsnoble larger question for you regarding the 'method' parameter. When Crema supported mixmax and TDC, it made sense to have this parameter. I also understand why removing this parameter makes sense if we remove mixmax since it's useless code. However, I think it might make sense for us to leave it in b/c it gives us the framework to add other methods in Crema. I remember at one point in time there was a discussion about putting some of Uri's other methods in Crema.

[wsnoble]

@wsnoble larger question for you regarding the 'method' parameter. When Crema supported mixmax and TDC, it made sense to have this parameter. I also understand why removing this parameter makes sense if we remove mixmax since it's useless code. However, I think it might make sense for us to leave it in b/c it gives us the framework to add other methods in Crema. I remember at one point in time there was a discussion about putting some of Uri's other methods in Crema.

If we ever need to reinstate this code, it will be available in the git history. I would prefer to avoid maintaining unused code.

[wsnoble]

Pushed a fix for the MSAmanda comment-line issue raised above (1ec1d73).

Problem: msamanda.py only recognized a single leading line starting with #version and skipped exactly one row. Real MS Amanda output can include several #-prefixed comment/settings lines before the header, which would break parsing (score columns wouldn't be found).

Fix:

• crema/parsers/msamanda.py: now loops while the line starts with #, counting however many comment lines precede the header, instead of checking for a single hardcoded #version prefix.
• crema/utils.py: generalized parse_psms_txt's skip_line parameter to accept an integer count of leading rows to skip (previously only 0 or 1 via a bool).

Note: I initially tried having parse_psms_txt auto-detect and skip any #-prefixed line itself, but that broke MSGF+ parsing — MSGF+'s real header column is literally named #SpecFile, so a blanket "skip lines starting with #" rule in the shared utility isn't safe. The comment-counting has to stay scoped to the MSAmanda parser, which is the only place that knows its header never starts with #.

Test: added a multi_comment_msamanda_csv fixture (3 leading # lines) and test_read_msamanda_multiple_comment_lines in tests/unit_tests/test_parsers.py. Confirmed it fails against the prior code and passes with the fix.

[bmx8177]

Alright, this looks good to me except for one thing. Specifically the CHANGELOG.md inside of the docs folder has not been updated to reflect the changes in this PR. One that is updated with something like "Remove mix-max dead code and --method CLI flag" we can merge.