Matchms

If you're working with mass spectrometry data in metabolomics or proteomics, this handles the grunt work of loading mzML, MGF, and MSP files, cleaning up messy metadata, and running spectral similarity calculations. The filtering pipeline is solid with 40+ functions for things like normalizing intensities, removing noise peaks, and harmonizing compound annotations. Spectral matching uses cosine similarity variants including modified cosine that accounts for precursor mass differences. The SpectrumProcessor lets you chain filters into reproducible workflows instead of writing the same boilerplate every time. It's Python native, open source, and frankly one of the better designed libraries in the computational mass spec space.