In the first part of this dissertation, I present here a method which utilizes a collection of conformations of a protein structure, accounting for the inherent property of protein flexibility in receptor-based pharmacophore modeling for virtual screening for inhibitors. This model, called a dynamic pharmacophore model, was developed using the conformations obtained by molecular dynamics simulation of the receptor. This ensemble of conformations was filtered through clustering methods to identify the most representative structures and was then subjected to GRID calculations to map-out the organic functional groups, complementary to the binding site. These features were finally overlaid to build a dynamic pharmacophore model. I tested several parameters like probe radii, exclusion volume radii and position, emergence of new binding site features and effect of simulation length, resulting in a defined set of rules for use in developing a dynamic model for maximum performance in a virtual screening protocol. Static pharmacophore models were also developed and their performance was compared with the dynamic models.

In the second part of this dissertation I present the simulation analysis of two different conformational forms of human Factor Xa (FXa), an essential protease in the blood coagulation cascade. These two different conformational forms of FXa (separate crystal structures) shows an open and closed S4 binding subsite. Through molecular dynamics and steered molecular dynamics studies I propose a hypothesis of the conformational diversity for FXa, according to which the two conformational isoforms are interchangeable, a property which may be important for FXa to interact with its various substrates in maintaining the blood coagulation pathway.