Biodegradability Evaluation of Chemical Structures

Related Papers:

A Biodegradability Evaluation and Simulation System (BESS),Applying and Discovering Knowledge of Biodegradation Pathways

Personell:

• Bill Punch , Intelligent Systems Lab, Dept. of Computer Science.
• Larry J. Forney , Department of Microbiology, University of Groningen, The Netherlands
• Arnold Patton , Intelligent Systems Lab, Dept. of Computer Science.
• Victor Miagkikh , Intelligent Systems Lab, Dept. of Computer Science.

Predicting the biodegradability of chemicals in the environment:

Over the past several decades, considerable progress has been made toward understanding the environmental fates of many naturally occurring and synthetic compounds. The important role of microorganisms in these transformation and mineralization processes has been clearly established. The ability to predict the biodegradability of chemicals in the environment is of increasing importance due to concerns about the persistence and toxicity of parent compounds and metabolites in the environment and the high cost of empirically assessing their fate. Thus, it is important to better understand basic principles that determine whether or not a given compound can be metabolized by microorganisms and the environmental and ecological factors that influence the rate and extent of metabolism in any particular ecosystem. Natural selection has resulted in the evolution of biochemical pathways able to metabolize naturally occurring compounds by microorganisms. The extent to which a synthetic chemical can serve as a substrate for various enzymes is dependent upon (i) the degree to which a synthetic chemical has structural features analogous to naturally occurring compounds; and (ii) whether synthetic compounds can be fortuitously transformed by extant enzymes with broad substrate specificity. For mineralization of a chemical to occur, several enzymes must act sequentially to transform or breakdown chemicals to molecules that enter pathways of intermediary metabolism. With expert knowledge of enzyme substrate specificity, sequential steps in the metabolism of naturally occurring compounds, and catabolic pathways, one identify structural features of synthetic chemicals that are likely to be substrates for enzyme(s) elaborated by microorganisms.

One can simulate the effects of these catabolic steps by representing a pathway as a "rule", which if invoked performs the same kind of structural modification as would the real pathway on the structure in question. The result of a sequence of such rule applications postulates a reasonable path for metabolism of a chemical and identifies metabolic intermediates that may accumulate. However, whether these predictions accurately reflect what is observed in nature is dependent on complex biological and environmental factors that are sometimes not well understood.

The approach we have taken is a collaborative effort between Environmental Scientists and Computer Scientists. To date, we have collected a working "book" of available pathways, and encoded these as rules into a simulator that predicts the biodegradability of the structure in question. Note a number of things. First, the biodegradability estimation simulates the breakdown of the structure based only on the structure, some environmental variables, and the application of the rule pathways. Second, the rules are actually more complicated than a typical rule-based system, encoding complicated pattern-matching algorithms to select where a pathway might be applied, and its effect. Third, we are now focusing on learning an organization of these rules, based on case performance and learning, to improve the accuracy of the simulator. Note that the latter is a balanced approach. We are not trying to learn the "rules" themselves, nor are we dictating how the rules will be invoked, but instead use established literature to establish the rules and then try to learn how to use them best.