Here, we report the development of EnzyMS, a Python-based pipeline for the analysis of high-resolution liquid chromatography-mass spectrometry (LC-MS) data specifically tailored for biocatalysis experiments. Applying EnzyMS to biocatalytic reactions carried out with variants of Fe(II)/α-ketoglutarate-dependent halogenase WelO5∗ on the antifungal macrolide soraphen A, we discovered reaction outcomes that had not been observable when using standard analysis software. Interestingly, we detected a previously unreported selective oxidative demethylation of soraphen A alongside the reported hydroxylations and chlorinations. Building on this finding, a computationally guided protein engineering approach allowed us to identify a WelO5∗ variant that exhibited a 3-fold improved demethylation performance by only creating and testing three predicted variants. In summary, we showcase the utility of the EnzyMS workflow and its potential to enable rapid detection of previously unobserved biocatalytic products and highlight the valuable synergies between data science pipelines and the computational design of enzymes.