2.3.0

ms2deepscore 2.3.0

Added

• New plotting functions for benchmarking #244

Changed

• Integrated new plotting functions in automated training pipeline #244
• Removed automatic storing of benchmarking scores #244
• Integrated loss calculation for validation loss and plots #244
• Validation loss uses all spectrum pairs instead of only 1 spectrum per inchikey #244
• Code refactoring and code linting.

Fixed

• Fingerprint type and number of bits specified in the settings is now correctly used in training and validation (before was set to default values in some instances) #244 and #251

Removed

• Removed version warning #249

Full Changelog: 2.2.0...2.3.0

2.2.0

What's Changed

• Adding case study notebooks by @niekdejonge in #229
• Switch linter (and linting style) from pylint + prospector to ruff #240
• Clearer documentation and naming to run training from existing splits #239
• empty cache to avoid memory leak by @florian-huber #243

2.1.0

What's Changed

• Update figure size by @niekdejonge in #214
• Make relu a setting by @niekdejonge in #228
• Update SettingsMS2Deepscore.py by @florian-huber in #222
• Small edits refactoring 232 by @niekdejonge in #235
• Cleaning refactoring 232 by @florian-huber in #233
• Improve inchikey pair selection and data generators by @niekdejonge in #232
• update zenodo link to latest model by @niekdejonge in #237

Full Changelog: 2.0.0...2.1.0

MS2Deepscore 2.0.0

MS2Deepscore 2.0 can train cross-ionmode models and was converted from tensorflow to pytorch. Which makes MS2Deepscore 2.0 incompatible with old models.

Full Changelog: 1.0.0...2.0.0

1.0.0

MS2DeepScore 1.0.0

Last version using Tensorflow. Next versions will be using PyTorch (and come with many changes and improvements!).

Added

• Added split_positive_and_negative_mode.py #148
• Added SettingMS2Deepscore #151
• Clearer Warnings when too little input spectra are used in data generator. #155

Changed

• Change the max oversampling rate to max_pairs_per_bin #148
• Made spectrum pair selection a lot simpler and fixed mistake #148
• Use DataGeneratorCherrypicked instead of DataGeneratorAllInchikeys in pipelines #148
• Removed M1 Chip compatibility which lead to faulty results depending on Tensorflow version #200

0.5.0

ms2deepscore 0.5.0

Added

• New DataGeneratorCherrypicked as an alternative to former data generators #145. This will work better for large datasets and also tried to counteract biases in the chemical similarity scores.
• Models can now be trained on selected metadata entries in addition to the spectrum peaks #128.
• New MetadataFeatureGenerator class to handle additional metadata more robustly #128
• Workflow scripts for training a new MS2DeepScore model #124. The ease of training MS2Deepscore models is improved, including standard settings and splitting validation and training data.

Changed

• In SiameseModel, the attributes are not passed as an argument but instead used by the class.
• Improved plotting functionality. Some additional plotting options were added and plots previously created in notebooks are now functions.
• Linting (code and imports) #145.

0.4.0

MS2DeepScore 0.4.0

Added

• Functions to cover the full pipeline of training a new model #129

Fixed

• Tensorflow issues when saving/loading models #123

Changed

• Random seed is now optional when fixed_set=True for the data generator #134
• load_model() functions now auto-detects if a model is multi_inputs or not
• Python version support was changed to 3.8, 3.9, 3.10 (other versions should still work but are not systematically tested)

0.3.0.1

Release

What's Changed

• make_tensorflow_macos_m1_chip_compatible by @niekdejonge in #110

Full Changelog: 0.3.0...0.3.0.1

0.3.1

MS2DeepScore 0.3.1

Minor changes to make tests work with new matchms (>=0.18.0). Older versions should work as well though. #120

0.3.0

ms2deepscore 0.3.0

Added

• Allow adding metadata to the network inputs, e.g. precursor-m/z using the additional_inputs parameter #115

Fixed

• Update test to work with Tensorflow 2.11 #114