Protein Structure Prediction Through Complex Modeling
Kevin Karplus, Ph.D.
A Member Since September 2002
As a professor of computer engineering at the University of California, Santa Cruz and also the Undergraduate and Graduate Director of Bioinformatics, Dr. Karplus is a leader at one of the strongest bioinformatics interdisciplinary programs in the United States. Although his particular expertise is in protein structure prediction and protein sequence analysis, he is well grounded in the fields of both mathematics and computer science. Upon graduating from Michigan State University with a Bachelor's degree in mathematics, he obtained a Master's degree in mathematics from Stanford University in California followed by a Ph.D. in computer science from the same institution.
After graduate school, Dr. Karplus focused his research on CAD tools for VLSI design, especially logic minimization before switching ultimately to biosequence analysis. He has worked extensively with hidden Markov models and Dirichlet mixtures.
Markov models represent stochastic processes, which generate random sequences of outcomes according to certain probabilities. A simple example of a stochastic process is a sequence of coin tosses, the outcomes being heads or tails. Markov models are used to analyze a wide variety of stochastic processes, from daily stock prices to the positions of genes in a chromosome.
A hidden Markov model is one in which one observes a sequence of results, but does not know the sequence of states the model went through to generate these results. Hidden Markov models are an important method of modeling a family of sequences or a common motif within a set of sequences.
In conjunction with Markov modeling, Dirichlet mixtures are employed to improve database search results by providing better estimates for which amino acids are possible in a given position, based on the ones that have been observed to occur.
Dr. Karplus also utilizes neural networks. A "neural network" is a classification or prediction program loosely inspired by the networks of biological neurons. Dr. Karplus uses neural nets primarily for predicting secondary structure and other local structural properties of proteins. In this application, neural nets have consistently outperformed other machine-learning methods.
In the summers of 1998, 2000, and 2002, Dr. Karplus served as a team leader for the biannual Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiment sponsored by the United States Department of Energy. This international community-wide experiment involves predicting the three-dimensional structures of target proteins using various computational tools. These target molecules are proteins currently under examination by experimental structural biologists. The experimentalists contribute the sequences of the amino acid chains of each protein to the entrants of the competition via the World Wide Web. CASP judges compare the predicted structures against final experimental structures, which are determined by different imaging techniques such as NMR spectroscopy and X-ray crystallography. The teams with the best results are invited to speak at the CASP conference and prepare papers for a special issue of Proteins: Strucutre, Function, and Genetics. Dr. Karplus's teams did well in all three years, with papers invited each time.
While devoted to his research, Dr. Karplus is also a dedicated bicyclist. He has made numerous contributions to improving bicycling for Santa Cruz residents through his personal devotion to alternative transportation. He has a life-long commitment to bicycling and has never owned a motor vehicle and does not even have a driver's license. He and his family, including his seven-year-old son, Abe, use only bicycles and public transportation.
Dr. Karplus has contributed his insights on many bicycle projects and road projects that impact cyclists in Santa Cruz. He is a certified Effective Cycling instructor and has taught bicycle engineering classes on campus. In addition, he has organized and continues to manage the Santa Cruz Bikes listserv, which provides a forum for bicyclists and others to participate in policy and technical discussion about projects that impact cyclists in his city and surrounding area.
<< Previous Next >>
[ View All Member Spotlights ]