RepoRT: Data Repository Enhances Metabolomics Research

Metabolomics is a central method to study health and disease and focusses on the analysis of small molecules - so called metabolites - from different biological specimens. The investigation of metabolites provides valuable insights into the biochemical pathways and metabolic activity occurring in cells, tissues or organisms. However, identifying and characterizing metabolites poses a significant challenge, due to the complexity and dynamics of these samples. Despite decades of research and numerous machine learning models, predicting small molecule retention times has remained a complex issue due to variations in compound structure and chromatographic systems. Retention time, a crucial parameter in metabolite identification, represents the time taken for a molecule to pass through a liquid chromatography column used for high-resolution separation of components in a sample. 

So far, new approaches for metabolite identification have focused on the development of data analysis strategies for mass spectrometric data. However, it becomes clear that certain limitations exist and orthogonal information such as retention times need to be leveraged, but only limited methods exist. In this study, the team of analytical chemists and bioinformaticians established RepoRT, a repository for metabolite retention times. Their goal: Enabling in future other bioinformatic and machine learning groups to develop algorithms for the enhanced prediction of retention times to aid metabolite identification. 

“This is the first repository delivering small molecule retention times as training data in a machine learning-ready fashion, which hopefully leads to a fast uptake of retention time prediction in metabolite identification workflows” says Michael Witting, the corresponding author of this article.