The Hitch-Hikers Guide to Computational Metabolomics

Since the 1970s, data processing units and computers have been sold alongside mass spectrometers. Over the past two decades, the concept of computational metabolomics has gained traction in scientific literature, with software capabilities evolving dramatically. These advancements have not only accelerated existing tasks but have also enabled researchers to address entirely new challenges. In the early days, manual inspection of chromatograms was standard practice. A major turning point for metabolite profiling came with xcms, one of the first open-source tools to implement a complete workflow from feature detection to univariate statistics. Over time, additional algorithms for feature detection, grouping, ion annotation, and many more, were developed. Research groups from all over the world have since contributed to the ever-expanding metaRbolomics ecosystem, with new tools and extensions available via CRAN and Bioconductor. With robust metabolite profiling workflows in place, the next challenge was metabolite identification and annotation. What once required days or weeks to identify a single metabolite can now be accelerated through computational approaches, enabling the annotation of (almost) all MS/MS spectra, faster than the measurements themselves. So, what's left to be done ? An increasing challenge was, and still is, interpreting the data in a biological and biomedical context. While the above workflow steps have been successfully automated, this final step still relies heavily on human expertise, intuition, and domain knowledge. Let's see what comes next. Social media: Don't panic! is also the recommendation when it comes to computational metabolomics, which does not only get stuff done faster, but opens new avenues to process and interpret metabolomics data.