Computational tools and workflows in metabolomics: An international survey highlights the opportunity for harmonisation through Galaxy

Over the past decade, metabolomics has transformed into a valuable scientific discipline, conducting comprehensive and (semi-)quantitative investigations of metabolism. Consequently, metabolomics is increasingly used across a wide range of research fields from basic biology to applied disciplines such as medicine, toxicology, environment and agriculture. Its expansion has been accompanied by a significant increase in the number of computational tools available to process and analyse metabolomics data. Such tools can provide automated and standardised operational pipelines for data pre-processing (e.g. chromatographic deconvolution or normalisation of data), univariate and multivariate statistical analysis, metabolite annotation and metabolic network modeling including deep learning strategies to detect new biological processes. Computational workflows are often complex and include a variety of tools and resources (e.g. scripting-language and version dependencies). For metabolomics to achieve its full potential in both the basic and applied sciences the accessibility, reporting, reproducibility and overall harmonisation of such computational tools and resources must be improved significantly. Only then can there be confidence that results obtained in one laboratory can be reproduced in another laboratory elsewhere across the globe. The development of standardised and reproducible computational workflows provides one route to achieving closer harmonisation. This is a logical evolution following the work of the Metabolomics Standards Initiative where one working group aimed to establish minimal reporting requirements for data processing (Goodacre et al. 2007 ).