els of scale from subcellular to systems. As part of the GENESIS project, Dr. Bower's research group has been developing software tools to facilitate access of modelers to the data on which their models depend and access of nonmodelers to model-based analysis of their systems. The GENESIS project also involves a significant educational component, which now forms the basis for many courses in computational neuroscience around the globe. Overall, the GENESIS project is intended to provide a new mechanism for scientific communication and collaboration involving both models and data. Dr. Bower's laboratory has also been involved in the development of silicon-based neural probes for large-scale multineuron recording procedures. These data are critical for the evaluation of network models of nervous-system function. Dr. Bower has a BS in zoology from Montana State University and a PhD in neurophysiology from the University of Wisconsin, Madison. He was a postdoctoral fellow at New York University and at the Marine Biological Laboratory in Woods Hole. He has been at California Institute of Technology since 1993.

Douglas Brutlag is director of the Bioinformatics Resource at Stanford University School of Medicine and professor of biochemistry and medicine at Stanford University. Dr. Brutlag's group works in functional genomics, structural genomics, and bioinformatics. They develop methods that can learn conserved structures, functions, features, and motifs from known protein and DNA sequences and use them to predict the function and structures of novel genes and proteins from the genomic efforts. The group uses statistical methods and machine learning to discover first principles of molecular and structural biology from known examples. They are also interested in predicting the interactions between ligands and proteins and between two interacting macromolecules and are actively studying the mechanisms of ligand-protein and protein-protein docking. Their research approach uses a variety of different representations of sequences and structures. Multiple representations of sequences include simple motif consensus sequence patterns, parametric representations, probabilistic techniques, graph theoretic approaches, and computer simulations. Much of the work consists of developing a new representation of a structure or a function of a macromolecule, applying the methods of machine learning to this representation, and then evaluating the accuracy of the method. The group has developed novel representations of sequence correlations that have predicted amino acid side-chain interactions that stabilize protein strands and helices. They have developed novel algorithms for aligning sequences that give insight into the secondary structure of proteins and developed novel methods for discovering both sequence and structural motifs in proteins that help establish