v2.1.0

What's Changed

This version now supports transfer leaning also for rust.

New Features

• Add decoy inclusion config switch by @GeorgWa in #746
• Use decoys from library by @GeorgWa in #780
• Add charge state prediction to PeptDeepPrediction step by @GeorgWa in #785

Bug fixes

• write precursors file also if no lfq results by @mschwoer in #773
• Skip charge prediction on fasta digest by @GeorgWa in #783
• Fix charge model training shape mismatch by @GeorgWa in #784
• Fix mbr config by @mschwoer in #782

Other Changes / unlabeled

• add error on non-specific digest by @mschwoer in #756
• rename dist search README by @mschwoer in #772
• Add conditional check for end-to-end tests based on commit age by @mschwoer in #775
• Remove obsolete features by @mschwoer in #770
• Remove two step classifier by @mschwoer in #771
• Remove neptune by @mschwoer in #778
• Skip quantselect test on specific macOS configurations to avoid seria… by @mschwoer in #781
• Add claude md by @mschwoer in #786
• Refactor DataFrame field access to use bracket notation for consistency by @mschwoer in #787
• remove hint in frontend by @mschwoer in #788

Full Changelog: v2.0.3...v2.1.0

v2.0.3

What's Changed

Bug fixes

• fix: ensure set_num_threads is called only once for NG computations by @mschwoer in #774

Other Changes / unlabeled

• Improve exit behaviour by @mschwoer in #757
• enhance output documentation with clarifications on identification an… by @mschwoer in #759
• Add type hints by @mschwoer in #752
• Simplify default yaml by @mschwoer in #758
• Improve UI inputs by @mschwoer in #760
• Fix UI tqdm by @mschwoer in #761
• Fix tests by @mschwoer in #769
• Quantselect_implementation by @vuductung in #741

Full Changelog: v2.0.2...v2.0.3

v2.0.2

What's Changed

Bug fixes

• set thread count for Rust backend in alphadia-search-rs by @mschwoer in #750

Other Changes / unlabeled

• adapt to pytest>=9 by @mschwoer in #749
• update README for additional guidance on conda installation by @vbrennsteiner in #751

Full Changelog: v2.0.1...v2.0.2

v2.0.1

What's Changed

Bugfixes

• Fix linux installer by @mschwoer in #745

Documentation

• Add Output documentation by @GeorgWa in #740
• Add AlphaDIA version to frozen config by @GeorgWa in #742

Other Changes / unlabeled

• feat: enhance pre-commit configuration with additional checks by @mschwoer in #735
• Fix motion styled version by @mschwoer in #737
• Run tests on cpu torch by @mschwoer in #738

Full Changelog: v2.0.0...v2.0.1

v2.0.0

What's Changed

This version now enabled using the new rust search backend.
It is superior in speed and performance, and recommended for Thermo and Sciex data (cf. the proteobench results for the performance-equivalent version v1.12.1).

However, some features are not yet available:

• transfer learning
• multiplexing
• processing Bruker files

It also comes with a series of breaking changes, both in the input (config) and in the output (column names).

New features

• Enable rust backend @mschwoer in #713
• Add parameter search functionality to gui by @mschwoer in #689

Breaking Changes

• The default output file format is now “parquet” instead of “tsv”
• Many output column names have been changed (see here)

[relevant for expert users only] The following config parameters have been renamed:

• fdr.competetive_scoring -> fdr.competitive_scoring
• search_initial.initial_num_candidates -> search_initial.num_candidate
• search_initial.initial_ms1_tolerance -> search_initial.ms1_tolerance
• search_initial.initial_ms2_tolerance -> search_initial.ms2_tolerance
• search_initial.initial_mobility_tolerance -> search_initial.mobility_tolerance
• search_initial.initial_rt_tolerance -> search_initial.rt_tolerance

[relevant for expert users only]In addition, some internal algorithmic parameters have been changed:
fdr.enable_nn_hyperparameter_tuning is now true by default
search_output.peptide_level_lfq and search_output.precursor_level_lfq are now true by default

• Add intensity to precursor table by @GeorgWa in #687
• Change config defaults by @mschwoer in #720
• rename competetive -> competitive by @mschwoer in #718
• Rename initial parameters in configuration files for consistency by @mschwoer in #719
• add semantic output names by @GeorgWa in #709
• Semantic stat keys by @GeorgWa in #726
• change stats by @GeorgWa in #729

Other Changes / unlabeled

• improve config comments by @mschwoer in #721
• fix test on windows: The key is that we exit the with block (which cl… by @mschwoer in #717
• log more system information including OS and Python details by @mschwoer in #716
• fix formatting by @mschwoer in #723
• output refactoring by @GeorgWa in #707
• Refactoring to isolate qouant output loading and building by @GeorgWa in #708
• bump version of rust backend by @mschwoer in #724
• Get rid of weird chars in windows GUI by @mschwoer in #693
• add rust dependency by @mschwoer in #725
• renaming by @mschwoer in #728
• fix: update intensity column mappings in output configuration by @mschwoer in #730
• fix: update output format for commitish_to_test in e2e testing workflow by @mschwoer in #731

unit/integration tests ran here: https://github.com/MannLabs/alphadia/actions/runs/18969706236
E2E tests ran here: https://github.com/MannLabs/alphadia/actions/runs/18970580325/job/54177201160

The linux installer was broken and has been removed, is fixed in this (functionally equivalent) release: https://github.com/MannLabs/alphadia/releases/tag/v2.0.1

Full Changelog: v1.12.2...v2.0.0

v1.12.2

What's Changed

Some minor bugfixes and improvements.

Bugfixes

• Use all cores in distributed search by @mschwoer in #694
• fix issue with frozen config when dumping numpy types by @mschwoer in #700
• explicitly set all dtypes by @mschwoer in #702
• Fix e2e slurm by @mschwoer in #704

Other changes

• Update citation by @mschwoer in #701
• allow for multiple config dicts by @mschwoer in #712
• Improve reuse quant by @mschwoer in #705

Full Changelog: v1.12.1...v1.12.2

v1.12.1

What's Changed

Bug fixes

• Add configurable option to do precursor/peptide level lfq by @mschwoer in #697

Full Changelog: v1.12.0...v1.12.1

v1.12.0

What's Changed

• Feature extraction (additional output) by @vuductung in #640

Other Changes / unlabeled

• Distributed search config fix by @vbrennsteiner in #633
• add first tests for quant_builder by @mschwoer in #641
• fix obvious bug in build_speclibflat_from_quant by @mschwoer in #646
• improve loading messages for execution engine by @mschwoer in #692
• Fully integrate NG backend by @mschwoer in #688

Full Changelog: v1.11.0...v1.12.0

v1.11.0

Algorithmic changes

• Fdr memopt by @mschwoer in #544
  => reduced memory requirements and runtime

These changes should avoid certain algorithmic failures:

• Filter fragments df by @mschwoer in #551
• Fix nonconverged estimators by @mschwoer in #681
• Correct tolerances in transfer lib req handler by @mschwoer in #684

Other relevant changes

• disable MBR step in default configuration by @mschwoer in #644
  => in the CLI only, this confused many people

Some expert features aiming for speed & reproducibity:

• test_numba_caching by @mschwoer in #543
• Allow fixed random seed 2 by @mschwoer in #670

Improved usability, logging and error messages:

• Improve usability by @mschwoer in #655
• Allow saving flat library by @mschwoer in #657
• fix inconsistent output paths by @mschwoer in #683
• Improve config handling by @mschwoer in #521
• Speed up -v and -h by @mschwoer in #539
• Invalid dia data error by @mschwoer in #552
• Wrap train test split error by @mschwoer in #553
• Re enable plotting by @mschwoer in #555

Non-functional changes / Refactoring

The backend has been refactored for modularity and readability.

• Slimmer utils by @mschwoer in #542
• Documentation by @mschwoer in #519
• Fix #546 by @jalew188 in #547
• small improvements of README.md by @mschwoer in #550
• Remove neptune from code by @mschwoer in #554
• Split plexscoring by @mschwoer in #556
• Refactor workflow paths by @mschwoer in #558
• Split features by @mschwoer in #559
• Split outputtransform by @mschwoer in #560
• Split workflow by @mschwoer in #561
• Split workflow ii by @mschwoer in #562
• More splitting by @mschwoer in #563
• update documentation configuration and add autosummary for modules by @mschwoer in #567
• add nb-clean pre-commit hook and apply by @mschwoer in #571
• Fix nbs by @mschwoer in #572
• Minor refactorings by @mschwoer in #573
• move quadrupole.py to plexscoring by @mschwoer in #574
• splitting_IV by @mschwoer in #575
• split_fdr_I by @mschwoer in #576
• split_fdr_II by @mschwoer in #577
• Lint fdr by @mschwoer in #578
• make_e2e_order_deterministic by @mschwoer in #583
• Remove unused methods from quadrupole by @mschwoer in #584
• split_fragcomp by @mschwoer in #585
• lint_fragcomp by @mschwoer in #586
• more_splitting_2 by @mschwoer in #587
• Split validate by @mschwoer in #588
• simplify validation by @mschwoer in #589
• Refactor numba by @mschwoer in #591
• Refactor workflow by @mschwoer in #592
• minor update in distributed search documentation by @vbrennsteiner in #605
• fix a test failing on windows runners by @mschwoer in #599
• Fix neptune in e2e by @mschwoer in #610
• Split containers by @mschwoer in #594
• Split peakgroups by @mschwoer in #595
• Split peakgroups ii by @mschwoer in #596
• Split peakgroups iii by @mschwoer in #597
• Split peakgroups iv by @mschwoer in #598
• inlinee _select_candidates, switch order of arguments by @mschwoer in #601
• Split peptidecentric i by @mschwoer in #602
• Split peptidecentric ii by @mschwoer in #603
• move out requantification_handler.pyi by @mschwoer in #604
• Split peptidecentric iv by @mschwoer in #606
• Split peptidecentric v by @mschwoer in #607
• Split peptidecentric vi by @mschwoer in #608
• Simplify init library by @mschwoer in #609
• split_peptidecentric_VII by @mschwoer in #611
• Misc refactorings by @mschwoer in #612
• Refactor managers i by @mschwoer in #613
• Refactor managers ii by @mschwoer in #614
• inline logic of disable_mobility_calibration() by @mschwoer in #615
• Refactor managers iv by @mschwoer in #616
• split optimizers by @mschwoer in #617
• Refactor managers vi by @mschwoer in #618
• Refactor managers vii by @mschwoer in #619
• Refactor data by @mschwoer in #620
• Misc refactorings ii by @mschwoer in #621
• Misc refactorings iii by @mschwoer in #622
• Refactor fdr manager by @mschwoer in #623
• Misc refactorings iv by @mschwoer in #624
• fix candidate index data type to int64 by @mschwoer in #634
• run also slow tests on every PR by @mschwoer in #627
• restrict output files to specific types in calc_metrics.py by @mschwoer in #642
• More timings by @mschwoer in #628
• Refactor search step by @mschwoer in #629
• remove previously removed classifier from fdr performance test by @mschwoer in #632
• Refactor raw data by @mschwoer in #635
• refactor extraction handler to prepare for multiple strategies by @mschwoer in #636
• Misc feedback by @mschwoer in #637
• reverting fd86f3f by @mschwoer in #645
• Add ng extraction handler by @mschwoer in #643
• Refactor plexscoring by @mschwoer in #647
• Introduce string constants by @mschwoer in #650
• Refactor calibration by @mschwoer in #651
• Refactor calibration ii by @mschwoer in #652
• Refactor calibration iii by @mschwoer in #653
• Refactor calibration iv by @mschwoer in #654
• Refactor calibration v by @mschwoer in #656
• Ng fixes by @mschwoer in #658
• disable norm and use batch norm by @mschwoer in #665
• Refactor search by @mschwoer in #660
• change classifier in to legacy_new_batching in fdr_test.py by @mschwoer in #667
• Refactor misc by @mschwoer in #662
• Improve gui 2 by @mschwoer in #674
• Dont repeat fdr for plotting 2 by @mschwoer in #671
• Improve logging by @mschwoer in #672
• Generify batch plan by @mschwoer in #685
• Misc changes by @mschwoer in #679
• manually bump alphabase version by @mschwoer in #682
• Dedicated ng defaults 2 by @mschwoer in #680

Full Changelog: v1.10.3...v1.11.0

v1.10.4

Bug fixes

add improved check for MS1 cycle #557

Full Changelog: v1.10.3...v1.10.4

Internal note: Hotfix release from https://github.com/MannLabs/alphadia/tree/release_1.10.4