Most efforts within the Human Genome Project are directed at discovering structural information, such as genomic maps and sequences. Therefore, most genome informatics efforts focus on managing and manipulating structural information. Yet knowledge of structure is of little value without an understanding of its functional manifestations.

The EcoCyc project is constructing a knowledge base that describes the genes and intermediary metabolism of E. coli. An associated graphical user interface provides biologists with facile access to this information. We have investigated high-fidelity representations for metabolic information, and we have developed visualizations that integrate metabolic and genomic information by linking automatic displays of metabolic pathways to a browser for the E. coli chromosome. Our group is also extending the capabilities of AI knowledge representation tools in support of the EcoCyc project. After describing EcoCyc, the talk will discuss challenging computational problems related to the metabolism, including knowledge acquisition, development of shared ontologies, design of novel biochemical pathways, and machine learning.