Study unveils unknown protein motifs (12/8/2007)

The task of determining protein structure and function is highly complex. Proteins are made of hundreds and sometimes thousands of atoms strung together that take on different forms and functions, and their sheer number and random variations make experimental approaches time- consuming and costly.

Drew Bryant '07 won the 2007 James S. Waters Creativity Award for his exceptional work in 3-D structural pattern matching for protein function prediction. His poster, "Integrating Novel Protein Structure Data for Improved Function Prediction Accuracy," also received first-place honors and the 2007 Jenessa Shapiro Award at the Rice Undergraduate Research Symposium.

For the past three years, Bryant has worked in Lydia Kavraki's Physical and Biological Computing Group with Brian Y. Chen '06, Rice statistical graduate student Viacheslav (Slava) Fofanov and his adviser, Marek Kimmel, professor of statistics, to develop computational techniques that compare functional information about well-studied proteins with uncharacterized proteins of similar geometric and chemical structure.

"Drew is among the brightest and most dedicated undergraduate students I've ever worked with, and he has brought valuable bioengineering perspectives to our projects," said Kavraki, the Noah Harding Professor of Computer Science and professor of bioengineering. "He used his biological insight to suggest properties that should be incorporated in the motif design and was a key player in the development of algorithmic tools that can identify those properties in the proteins stored in the Protein Data Bank."

To distribute the calculations and consider all proteins in the Protein Data Bank, Bryant and Chen orchestrated searches using the 600 processors of the Rice ADA cluster. Then they efficiently identified matches between motifs and protein targets with high sensitivity and specificity. While with Kavraki's group, Bryant also built a Web server for making the protein prediction software available to the scientific community.

"We wanted to develop a rigorous computational framework that allows one to design, test and tweak motifs quickly so that the optimal representation of the protein's function can be found," said Bryant. "I hope to use these statistics to understand the problem of predicting protein function more accurately."

Bryant's work has been part of a larger project in Kavraki's laboratory that is supported partially by a National Science Foundation subcontract from Baylor College of Medicine, with principal investigator Olivier Lichtarge, Baylor professor of molecular and human genetics and of biochemistry and molecular biology. Initially, a pipeline called the Match Augmentation with Statistical Hypothesis Testing (MASH) was built. Once the pipeline was fully operational in 2006, it was further developed and used to refine motifs provided by Lichtarge and his group. The product of this work led to the development of geometric sieving - a method that refines candidate motifs into optimized motifs with maximal geometric uniqueness from all known protein structures.

This year, Bryant's accumulative work has received tremendous attention. He is a major contributor to two papers, "Cavity Scaling: Automated Refinement of Cavity-Aware Motifs in Protein Function and Prediction," published in August 2006 by Imperial College Press; and "The MASH Pipeline for Protein Function and Prediction and an Algorithm for the Geometric Refinement of 3-D Motifs," published by the Journal of Computational Biology.

This past summer while working at Genentech in San Francisco, Bryant completed a third paper that investigates different conformations of the same protein structure to define motifs. He will continue his graduate studies with Kavraki.

Note: This story has been adapted from a news release issued by Rice University

Post Comments: