DNA Recognition by X-ray & statistical mechanics


 

My research philosophy is to bring a physical, problem-ori­ented ap­proach to the acquisition and interpretation of structural biomolecular information. I view biological molecules as amazing machines that have reached the ultimate miniaturization possible in a universe composed of atoms and molecules; my approach is to pick interesting examples where it appears possible to develop understanding of how they work.
Structure determination is central to this effort and the laboratory is well equipped for macromolecular X-ray crystallography. Local facilities include four area detectors, crystallization robots, computers and graphics as well as facilities for expression/synthesis and purification of proteins and oligonucleotides. Additional, external resources include synchrotron X-ray sources.

Brownian motion dominates the dynamics of proteins and nucleic acids; hence understanding the connection between structure and function requires the application of concepts of statistical mechanics and computation. Here, the approach is to take problems suggested by examination of structures and perform molecular dynamics simulations followed by calculations of free energies and related values. These have required ongoing, collaborative efforts (primarily with Robert Swendsen); directed towards the development of computational methods to facilitate these calculations. Resources for these computations include the Pittsburgh Supercomputing Center as well as local workstations and a small Beowulf cluster we have constructed.

Structure-related information is accumulating at an exponentially growing rate; informatics is increasing necessary to make effective use of it. Initially, we began developing methods (in collaboration with Bruce Buchanan); for the use of crystallization and biochemical information to assist in the determination of the appropriate crystallization conditions for newly investigated macromolecules. Techniques used here include Bayesian case reconstruction, machine learning, clustering, and machine vision. More recently, we have also begun applying these methods to the automated analysis of structural databases, such as the Protein Data Base. These include for the automatic identification of structural motifs as for automatic criteria for the evaluation of crystallographic quality. An important theme running through this work is that the informatics tools force one to "systematize one's belief system." Doing so has proven both enlightening and fun.

 


 

Education

PhD 1974, Massachusetts Institute of Technology

Postdoctoral Training

California Institute of Technology, Laboratory of Dr. Richard Dickerson


Department of Bioological Sciences
University of Pittsburgh
312 Clapp Hall
4249 Fifth Ave.
Pittsburgh, PA 15260

Phone: (412) 624-8778
Fax: (412) 624-8109

E-mail: jmr@jmr3.xtal.pitt.edu

Visit website