The Universityof Minesota is awarded a grant to further develop a prototype computational system that predicts metabolic pathways for any given compound. The current system predicts all possible pathways based on metabolic rules; hundreds of pathway branches are generated. To use this effectively to predict the fate of chemicals in the environment, reasoning capabilities will need to be added. The reasoning tools, developed during the research proposed here, will select likely pathways, and be a platform for discerning different chemical fates in different environments, as needed by regulators and industry. The research project is entirely in the public domain and will connect to many partners worldwide. It will be synergistic with the efforts of Proctor & Gamble, Brussels and the ALARM project of the 6th European Union Framework Program.
The Pathway Predictive System (PPS) is based on 250 metabolic rules. PPS rules simulate virtually all plausible biochemical pathways for a compound, based on comparison with human experts, known pathways and bench experimental verification. To move toward a true depiction of environmental fate, the first standard condition to be simulated will be the aerobic environment, with moderate moisture and salinity, no competing chemical toxicants, and at 25oC. A comprehensive approach will be used to prioritize biotransformation rules. First, human experts will rate a rule as applying under those conditions on a 1-5 scale: 1) Highly likely; 2) Likely; 3) Neutral; 4) Unlikely; 5) Highly unlikely. This will provide an ordinal ranking of reactions. Expert knowledge will be augmented using thermodynamic techniques. Preliminary data suggests that thermodynamics can minimally select against futile metabolic cycles and thermodynamically unfavorable isomerization reactions. Rapidly-growing microbial genomic data will also be studied to improve biotransformation prioritization. The project will quantitatively assess the prevalence of genes and enzymes that participate in reactions covered by current PPS rules. There will be a further assessment of whether those genes and enzymes are found in bacteria specific to certain environments, such as aerobes, anaerobes, and halophiles. The system will be developed using the resources of the University of Minnesota Supercomputing Institute and be publicly mirrored in Europe, for use by participants in the ALARM program and other clients. The rules and use of the decision-making software will be freely available to all users on the web. The system will also be used educationally in University classes throughout the nation via a network of collaborating academicians. This will be used to teach students chemistry, metabolic biochemistry, genomics and environmental sciences, and help us validate the pathway priority prediction system. All of the results will be made available to the scientific community through our website.
