Ph: (217) 333 1630, FAX: (217) 244 6615, E-Mail: biophysics@life.uiuc.edu
Director: Martin Gruebele Administrative Coordinator: Cindy Dodds
|
Quick Links:
UIUC Student Links:
Grad College Links:
School of MCB Links:
Other Program Links:
Outside Links:
|
Faculty Profiles
Current Research My research focuses on establishing a statistical mechanical framework to study protein structure prediction and protein folding. Specifically I am interested in developing optimized energy functions for protein tertiary structure prediction (1-3), and then using them along with other computational tools and approaches to compare across various genomes the structure and function of proteins involved in key metabolic pathways. A more recent research direction in my efforts to understand the underlying chemical and physical principles of structural genomics includes the study of variation in the physical properties of DNA between coding and regulatory regions. Protein Structure Prediction With the publication and preliminary analysis of over 21 complete genomic sequences, it is now clear that there are only a limited number of distinct protein folds. Since the template structures of roughly one-third of these folds are known, one aspect of our work has been to develop statistical mechanical methods to evaluate if a protein sequence has properties compatible with any of these templates (4). These properties are incorporated in dynamic programming and hidden markoff algorithms that are used to carry out the threading or aligning of a sequence onto a template structure. The energy functions based on these properties are also used to identify stable folding subunits (foldons) which are important in understanding the mechanism of folding and for many aspects of design (5, 6). To predict the structure of proteins with novel folds, we have developed an ab initio method with interactions based locally on association of the sequence with a database of structural patterns and globally on agreement with simple pair correlation functions typical of globular proteins (2, 7). The lowest energy basin of conformations satisfying this associative memory hamiltonian are sampled using molecular dynamics with simulated annealing. The energy functions used in both prediction approaches are optimized with respect to simple principles that distinguish proteins of Nature from random heteropolymers of amino acids (8,9). Elucidating these principles and determining the appropriate reaction coordinates to describe the folding process are major challenges to predicting the structure of a protein from its amino acid sequences. Using low resolution lattice models of proteins and straightforward thermodynamic arguments, we showed one such fundamental principle is that the energy landscape of a foldable protein must be funnel-like (10). Brownian Dynamics in Biological Systems In a mesoscopic treatment, the Fokker-Planck equation and its adjoint can be used to study the dynamics of biological reactions involving diffusive barrier crossing (11, 12). Examples of such reactions are the association of the ends or fragments of a biopolymer, protein folding, and the conformational transitions induced in biopolymers by the application of a weak constant force in atomic force microscopy experiments. We have recently used estimates of the first-passage time derived from the adjoint equation to compare the conformational barriers estimated from an AFM experiment on polysaccarides with those resulting from quantum chemical calculations. |
