Add InstaNovo and InstaNovo+ import support

[BioGeek]

InstaNovo is a de novo peptide sequencing tool.

This PR adds support for normalized CSV output from InstaNovo v1.2.2 in three modes:

• InstaNovo: de novo peptide sequencing with the transformer-based InstaNovo model.
• InstaNovo+: de novo peptide sequencing with the diffusion-based InstaNovo+ model.
• InstaNovo with refinement: InstaNovo predictions refined with InstaNovo+.

This PR forms the basis for a bunch of related PRs within the CompOmics ecosystem.

Changes

• Adds built-in advocates for:
  • InstaNovo
  • InstaNovo+
  • InstaNovo with refinement
• Adds identification-file readers for:
  • .instanovo.csv
  • .instanovoplus.csv
  • .instanovo.refined.csv
• Adds support for normalized InstaNovo v1.2.2 CSV columns.
• Adds support for the default UniMod annotations emitted by InstaNovo's default residue configuration.
• Adds sample-row based tests for InstaNovo, InstaNovo+, and refined output.
• Improves spectrum-title matching for real MGF/mzML titles.
• Handles whitespace-padded precursor charges and skips rows with invalid charges.
• Adds shared InstaNovo parameter classes used by SearchGUI advanced settings.

Supporting Fixes

Shared fixes needed for the InstaNovo de novo end-to-end flow (SearchGUI → PeptideShaker):

• Declares the ThermoRawFileParser mzML output ending as .mzML (the canonical extension the tool writes) so converted spectra are found on case-sensitive file systems.
• Aligns log4j-api to 2.25.4 to match the bumped log4j-core, fixing a runtime NoSuchFieldError.
• Suppresses (logs but does not pop a dialog for) the benign Nimbus look-and-feel ClassCastException (ColorUIResource cannot be cast to Boolean) thrown while building chart popup menus on recent JDKs.

Implementation Notes

• The refined workflow is represented separately from standalone InstaNovo+ so downstream tools can distinguish both result types.
• Refined output reports software metadata for the refined workflow and the underlying InstaNovo/InstaNovo+ tools.
• Spectrum title lookup is cached per spectrum file during CSV import.

Tests

Ran:

mvn -Dtest=com.compomics.util.test.experiment.io.identifications.TestInstaNovoIdfileReader,com.compomics.util.test.parameters.identification.tool_specific.TestInstaNovoParameters test
mvn -DskipTests -Dmaven.javadoc.skip=true install

Covered by tests:

• InstaNovo v1.2.2 CSV parsing.
• InstaNovo+ v1.2.2 CSV parsing.
• InstaNovo with refinement v1.2.2 CSV parsing.
• default InstaNovo UniMod modification mapping.
• realistic non-numeric spectrum titles with scan= tokens.
• invalid and whitespace-padded precursor charges.

Merge Order

Recommended order:

1. This PR
2. Add SearchGUI support for InstaNovo and InstaNovo+ searchgui#387
3. Add PeptideShaker import UI support for InstaNovo result files peptide-shaker#570
4. Add DeNovoGUI compatibility APIs to Utilities #74
5. Add InstaNovo support to DeNovoGUI denovogui#53
6. Update SearchGUI recipe to 4.3.17 bioconda/bioconda-recipes#66604
7. Update PeptideShaker recipe to 3.0.13 bioconda/bioconda-recipes#66603
8. Expose InstaNovo workflows in online input form barsnes-group/peptide-shaker-online#30
9. Bump PeptideShaker wrappers for InstaNovo support galaxyproteomics/tools-galaxyp#828

[BioGeek]

Hello @hbarsnes,

Any feedback on this PR and the related PRs? I'm happy to make any requested changes or answer questions if anything needs clarification.

[hbarsnes]

Hi @BioGeek,

I'm afraid that all of the projects affected by your pull requests are no longer in active development and have not been so for quite some time. We sadly therefore do not have the time and resources to verify the changes, correct any potential conflicts or reply to new issues if there would be any.

I also think that the number of users of these tools have been rapidly diminishing over the last couple of years (largely due to the lack of updates and improvements). Hence, there would probably not be a lot to gain by adding new features at this point.

This situation is not expected to change, and at some point the gap compared to similar tools will simply be too big for it to make sense to put in the required effort.