matchms 0.31.0

Added

• New FlashSimilarity score class [#827](https://github.com/matchms/matchms/pull/8267. This conceptually follows the Flash Entropy proposed by Li & Fiehn (2023) and allows much faster computation than, for instance, CosineGreedy or ModifiedCosine. It is possible to choose different score_type (can be spectral entropy or cosine) as well as matching_mode (can be fragment, neutral_loss or hybrid). To compute modified cosine scores, for instance, this would be cosine together with hybrid).
• New BlinkCosine similarity score that is largely following the BLINK score proposed by Harwood et al. (2023), representing a fast-to-compute approximation of the Cosine score between MS/MS spectra #829.
• New add_precursor_formula filter #826.
• New remove_peaks_relative_to_precursor_mz filter #833.

Changed

• Added Min-Max scaling to normalize intensities filter #830.

Full Changelog: 0.30.2...0.31.0

What's Changed

• Fixed metadata export by @hechth in #814
• prepare release 0.30.2 by @hechth in #815

Full Changelog: 0.30.1...0.30.2

What's Changed

• Fix msp import set key as numpy scalar instead of value by @julianpollmann in #807
• Bump requests from 2.32.3 to 2.32.4 by @dependabot in #808
• matchms fails when reading spectra where abundance is in scientific notation by @hechth in #809
• Bump urllib3 from 2.4.0 to 2.5.0 in #811

Full Changelog: 0.30.0...0.30.1

What's Changed

• add support for python 3.13 by @julianpollmann in #804
• upgrade numPy from >1.24 to >=2.0.0

Full Changelog: 0.29.0...0.30.0

What's Changed

Added

• Implemented preliminary mzSpecLib export #757
• added BinnedEmbeddingSimilarity and BaseEmbeddingSimilarity Classes #749
• added Fingerprints Class to compute and store inchikey-fingerprint mapping for a list of spectra #717
• some reference spectra were added #781

Changed

• compound_name is now always the first attribute to be written for each spectrum #762
• added option to use different peak separators for msp export #762
• filtering: cloning of Spectra is now optional in filtering and disabled in SpectrumProcessor.
  Enable cloning for the use of ProcessingReport with create_report = True #754
• importing now supports pathlib.Path #738
• exporting: empty spectra are saved as empty file instead of not saving at all #722
• exporting: save_as_mgf now supports write and append mode #741
• importing: load_mgf supports StringIO #745
• importing: fixed bug load_from_usi API call #759
• filtering: normalize_intensities will now set intensities to 0 instead of None when peak intensities are 0 #750
• Support GOLM style MSP files #763
• omit prospector, isort, black in favor of ruff #790

New Contributors

• @wasimsandhu made their first contribution in #722
• @mvisani made their first contribution in #745
• @roman-bushuiev made their first contribution in #738
• @acquayefrank made their first contribution in #741
• @oolonek made their first contribution in #748
• @Knerten0815 made their first contribution in #750
• @LucaCappelletti94 made their first contribution in #781

Full Changelog: 0.28.2...0.29

What's Changed

• Allow numpy 1.x and 2.x by @florian-huber in #715

Full Changelog: 0.28.1...0.28.2

Includes changes to the readme for PyPI compatability.

Full Changelog: 0.28.0...0.28.1

What's Changed

• Increased Test Coverage by @julianpollmann in #701
• Rename CudaMS -> SimMS, tweak description a bit by @tornikeo in #703
• Update utils.py by @niekdejonge in #705
• Updated matchms dependencies by @hechth in #709
• IndexError in .matrix when all scores are 0 by @tornikeo in #702
• add logging for writing spectra to file by @florian-huber in #645
• Enable metadata exporting with tab separators by @hechth in #712

Full Changelog: 0.27.0...0.28.0

[0.27.0] -2024-07-10

Changed

• Avoid using unstable sorting while sorting collected matching peaks #636.
• Losses will no longer be stored as part of a Spectrum object, but will be computed on the fly (using spectrum.losses or spectrum.compute_losses(loss_mz_from, loss_mz_to))#681
• Jaccard/Tanimoto @njit/numba-based similarity functions were replaced by 10-50x faster numpy matrix multiplications #638.
• Dependencies were updated to allow newer numpy and numba versions 691.
• Renamed method names and parameters to align spectrums -> spectra
• Python support changed from 3.8 - 3.11 to 3.9 to 3.12, and dependency versions were updated 640.

Removed

• add_losses() filter was removed. Losses will no longer be stored as part of a Spectrum object, but will be computed on the fly #681.

Fixed

• Remove empty spectra before exporting to file #686.
• Name position in mirror plots #678.

What's Changed

• Derive formula from smiles by @niekdejonge in #661
• Add check filters handle none by @niekdejonge in #665
• fix merge mistake by @niekdejonge in #677
• Add require number of peaks below maximum by @niekdejonge in #664

Full Changelog: 0.26.3...0.26.4