A Qualitative Biochemistry and Its Application to the Regulation of the Tryptophan Operon

This article is concerned with the general question of how to represent biological knowledge in computers such that it may be used in multiple problem solving tasks. In particular, I present a model of a bacterial gene regulation system that is used by a program that simulates gene regulation experiments , and by a second program that formulates hypotheses to account for errors in predicted experiment outcomes. This article focuses on the issues of representation and simulation; for more information on the hypothesis formation task see (Karp, 1989; Karp, 1990). The bacterial gene regulation system of interest is the tryptophan (trp) operon of E. coli (Yanofsky, 1981). The genes that it contains code for enzymes that synthesize the amino acid tryptophan. My model of the trp oper-on—called GENSIM (genetic simulator)—describes the biochemical reactions that determine when the genes within the operon are expressed and when they are not, the reactions by which the genes direct the synthesis of the biosynthetic enzymes (transcription and translation), and the reactions cat-alyzed by these enzymes. Therefore my modeling techniques are specifically designed to represent enzymatically-catalyzed biochemical reactions whose substrates include macromolecules with complex internal structures, such as DNA and RNA. These techniques address such issues as: How might we represent the attributes and the structures of the objects that make up the trp operon? What is a suitable ontology for these objects—an appropriate level of abstraction at which to model them? How might we describe a gene regulation experiment, and how can we maintain a library of known experiments? How might we represent known biochemical reactions? How can we design a simulation program that predicts the outcome of a gene regulation experiment by correctly and efficiently simulating every reaction that occurs in that experiment, and only those reactions? GENSIM embodies a qualitative biochemistry because it provides a framework for representing knowledge of biochemistry, and for performing qualitative simulations of biochemical systems. The specific features of this qualitative biochemistry are as follows. I employ frames to represent biochemical objects that correspond to homogeneous populations of molecules. This representation describes the decomposition of complex objects into their component parts. I use frame knowledge bases to represent the objects present in the initial conditions of different experiments. Section 2 describes how instance frames represent chemical objects in simulation knowledge bases; Section 3 describes how class frames are used to represent general classes of biochemical objects, and presents …