Default to trypsin/P so results don't change under mzLib #1186 (needs mzLib #1186)

[trishorts]

Project of origin: none — standalone fix (companion to smith-chem-wisc/mzLib#1186)

Draft on the fork until the mzLib release that carries smith-chem-wisc/mzLib#1186. At that point:

1. bump the mzLib pin in the six csproj files;
2. open this on smith, in the same pin-bump PR as fork Read PSI-MOD formal charges from mzLib's PsiModFormalCharges.tsv #27, Quantify a design with a numbering gap, with a warning in the log and results.txt, instead of refusing it #29 and Expect a sequence in two indistinguishable proteins to quantify their group (needs mzLib #1463) #30;
3. add the labels, reviewers and board.

This draft is based on smith-chem-wisc#2905's branch, so the diff shows only this change. Retarget it to master once smith-chem-wisc#2905 merges. Closes smith-chem-wisc#2750 when it lands on smith.

Why

smith-chem-wisc/mzLib#1186 restores the protease naming every other search engine uses:

• bare trypsin is the proline-restricted enzyme, K[P]|,R[P]|;
• trypsin/P cleaves before proline, K|,R|.

Bare trypsin has meant the unrestricted enzyme since at least 2022, and that is what MetaMorpheus searches with by default. Against a pack of smith-chem-wisc/mzLib#1186, MetaMorpheus master fails 77 of 2,910 tests:

• 25 change because bare trypsin now restricts proline: different peptides, PSM counts and q-values.
  • 3 get it from MetaMorpheus's own default.
  • 22 name trypsin, or take mzLib's DigestionParams default.
• 2 involve the retired X|P names:
  • one task file fails to load (fixed in Load task files that name a protease by a historical spelling smith-chem-wisc/MetaMorpheus#2905);
  • one test round-trips a retired name (updated here).
• 50 are knock-on failures from a test flag that leaks once an earlier test fails (fixed in Reset the q-value override in tests even when an assertion fails smith-chem-wisc/MetaMorpheus#2906).

The decision for MetaMorpheus is to keep today's results: default to trypsin/P and name it wherever a test's expected values were measured with it.

What this changes

• CommonParameters.DefaultProteaseName = "trypsin/P":
  • CommonParameters uses it when no digestion parameters are given.
  • The search, calibration, GPTMD, glyco and crosslink windows and the file-specific parameters window start on it.
  • The Arg-C and semi-trypsin recommended-settings switches select it.
• GUI recommended-settings cases:
  • The chymotrypsin cases now match chymotrypsin and chymotrypsin/P. They matched the parenthetical names mzLib make protease and rnase tsv embedded resources smith-chem-wisc/mzLib#1005 replaced in February 2026, so they never fired on a current mzLib.
  • The elastase case starts firing again by itself, since Make the proline suffix follow the community convention (|P was inverted) smith-chem-wisc/mzLib#1186 renames elastase|P back to elastase.
• Tests:
  • Tests whose expected values were measured with the unrestricted enzyme now name trypsin/P: 18 lines in 8 test files.
  • So do six settings files. One is TestData/SmallCalibratible_Yeast.toml, a file-specific settings file that several searches pick up. Its bare trypsin overrode the task's protease, which is why TestPepFilteringFewerThan100Psms changed even with the new default.
  • Tests that pass under either rule keep their spelling.
• DigestionAgentNameTests.ASavedTaskReloadsWithItsProtease round-trips chymotrypsin and trypsin/P instead of the retired chymotrypsin|P.

Not changed

Saved task files. A saved task file that says Protease = "trypsin" now loads as the restricted enzyme. Every such file was saved meaning the unrestricted one, so re-running it changes its results. smith-chem-wisc/mzLib#1186 documents this as its one deliberate break. Nothing in this PR handles it; whether MetaMorpheus should is an open decision (see smith-chem-wisc#2750).

Verified

Against a local pack of smith-chem-wisc/mzLib#1186 at 5b424f6a2 (its later head adds only an error-message change), with smith-chem-wisc#2905, smith-chem-wisc#2906 and this change, the full suite (filter Category!=ExternalService, as mzLib's integration job runs it) gives 2912 passed, 0 failed, 1 skipped.

🤖 Generated with Claude Code

https://claude.ai/code/session_013NnV5rYxLKC7SNfcgT1uEV