Unlocking the Molecular Secrets of the Human Body: Exploring Metabolites with Machine Learning

Scientists use machine-learning to gain a new perspective on small molecules

A new machine-learning model will assist scientists in identifying small molecules. It has applications in medicine and drug discovery, as well as environmental chemistry. Researchers at Aalto University in Finland and University of Luxembourg developed the machine learning model, which was then trained using data from dozens of labs to become the most accurate tool for identifying small molecule.

The human body is made up of thousands of small molecules known as metabolites. These molecules transport energy throughout the body and communicate cellular information. Because they are small, it is difficult to differentiate metabolites in a blood analysis. However, understanding how exercise, nutrition and alcohol consumption affects well-being requires identifying these molecules.

Mass spectrometry is used to identify metabolites by analysing their retention time and mass using a liquid chromatography separation technique. This technique separates metabolites first by running the sample down a column. The result is different flow rates, or retention times through the measurement device.