Biophysics of Molecular Recognition

Complex physical processes such as protein folding, protein-DNA and protein-protein interactions are central for functioning of any biological system from a simple virus to a mammalian organism. Molecular biology has been studying these processes for more than 30 years, and structural biology provides pictures of molecular interfaces at atomic resolution. However, the physics of molecular recognition is anything but understood. How does a transcription factor recognize its site on the DNA from millions of other sites (and does this in a matter of seconds)? How does each kinase recognize its specific substrate protein, while other proteins are 80% identical? Why do DNA-binding domains fold only when they bind DNA?

Our goal is to understand the underlying physical mechanisms of specific molecular recognition and develop computational tools to study rapidly growing genomic information. For example, for protein-DNA interactions we aim at development of a program able to predict DNA sites recognized by a protein given a sequence (and a structure) of the protein. This way hundred of novel transcription factors can be assigned to the genes they regulate.

The main tools we use are computer simulation and bioinformatic analysis. In the simulations, we construct biophysical models of how a protein folds, binds DNA or other proteins. Bioinformatics allows us to study proteins without making experiments: natural evolution have tried numerous mutations and provided us with the record of successful experiments. We develop bioinformatics tools to read these records and understand roles of individual amino acids or nucleotides in folding, binding and specific recognition.

Physics of Biological Networks

In molecular biology, biological systems used to be studied by reducing them to elements, pathways, and reactions simple enough to be well characterized in terms of the interactions between specific proteins, nucleic acids, and small molecules. A great deal of understanding of individual biological mechanisms and processes has been gained this way. However, it has become evident that focusing on individual pathways and processes does not lead us to a broader picture of the whole cell, tissue, or organ. Our goal is to develop system-level understanding of biological circuits, their structure and dynamics. Can we find novel structural elements (complexes, pathways, and modules) in the known biological networks using genomic and proteomic data? What's the origin of incredible robustness of biological networks and how do they combine robustness with adaptability? How do biological networks evolve?

Here again we combine bioinformatic techniques with computer simulations. In simulations, we develop models of biological networks, study their dynamics and evolution. Bioinformatics, in turn, helps us to mine proteomic and genomic data to discover new features of biological networks, understand how they work and evolve.

Our research has many practical applications in the biomedical sciences and beyond. Discovering principles used by nature to design robust and adaptive biological networks, can aid in the engineering of artificial biological and non-biological networks with such properties. Understanding tolerances and failures of biological networks will help to identify genes and proteins that are likely to be linked to diseases and that are potential targets of drug development.

Selected Publications:

Mirny LA, Gelfand MS. Using orthologous and paralogous proteins to identify specificity determining residues. Genome Biol. (2002); 3(3)

Mirny, L.A. and Gelfand M.S. Structural analysis of conserved base pairs in protein-DNA complexes. Nucl. Acids. Res. (2002) 30: 1704-1711.

Mirny, L.A. and Shakhnovich,E.I. Evolutionary conservation of the folding nucleus. Journal of Molecular Biology (2001) 308:123-129