Grant: Project Grant

Small molecule compounds such as drugs, nutrients and pollutants affect our health in many ways. Nutrients are needed to sustain life, drugs are needed to cure us of illnesses, while pollutants can kill us. Millions of these compounds exist. The complete collection of these compounds is called the "metabolome" -- derived from the word metabolite (a substance produced by metabolism). The study of what these metabolites are and how they affect us is called metabolomics. Metabolomics is a rapidly emerging field of science that uses complex analytical instruments such as mass spectrometers to identify and quantify 100s to 1000s of chemicals in the body. Unfortunately, only about 5-10% of these compounds can be positively identified. Furthermore, of the relatively few chemicals we can identify, we often don't know where they come from, we can't tell if they are good or bad and we rarely know how they act on our cells or modify our genes. Given the current state of knowledge about the metabolome and the limited experimental capabilities we have in metabolomics, new approaches must be found to advance the field. One promising route is to use computers. In particular, by using some of the latest techniques in artificial intelligence, text mining and machine learning, it should be possible to make metabolomics much faster and far better. My plan is to develop novel computer programs to predict chemically feasible structures, to calculate accurate spectral patterns, to vastly increase the number of compounds we can identify and to automatically infer the biological roles that these previously unidentified or little-known metabolites have in the human body. These kinds of computational tools will not only transform metabolomics, they will accelerate the discovery of new drugs, help identify the underlying chemical causes of diseases, improve medical diagnostics and assist with predicting those at risk of developing diseases such as cancer, heart disease and diabetes.