April 6, 2006
The rapid proliferation of structural molecular data of the living cell in recent decades obviates the need for adequate computational analytics to represent, manipulate and analyze these in ways that maximize their scientific utility.
Biologically active molecules (proteins, nucleic acids, and their combination in macromolecular complexes) have distinct three-dimensional structures that match and fit, and thereby determine their behavior and interactions with other molecules within their native cellular environment.
Moreover, in order to successfully apply molecular structural information in areas such as drug discovery, and disease therapy, one must also be able to take into account energetic factors such as electrostatic potentials and hydrophobicity for more comprehensive biomolecular interactions.
Complicating the analytics still further is the fact that most molecules are flexible, being able to adopt a number of different conformations that are of similar energy.
For this reason, the rigid-model approximations once used to represent molecules are no longer elaborate enough to provide adequate predictions of chemical and physiological activity.
Advances in the computational power, on the other hand, now allows less restrictive and more dynamic models of biomolecular complexes to be employed in computations and simulations.
In this talk I shall cover several geometric and signal processing algorithms that are required to support both the elucidation of atomic and quasi-atomic models of macromolecules from electron microscopy, and the validation of flexible and dynamic models of bio-molecular and macromolecular interactions.
Chandrajit Bajaj (University of Texas, USA)