Research Projects on Hydra Multiscale Molecular Dynamics and its Applications in Nano–Fluid Dynamics and Computer–Assisted Drug Design

Molecular dynamics (MD) solves a system of ordinary differential equations governing the motion of the particles (atoms) in a system [2]. MD is very useful in many applications. In a biomolecular system such as solvated proteins, for example, one can use MD to study the coordinated motion of the side chains, the closing and opening of certain binding domains, the diffusion of small molecules inside the channels within the proteins, or the interaction of ligand (drug) molecules and proteins [12, 29]. These studies are crucial in providing deep insights on a molecular level as to why certain molecules are useful for regulating protein function, thereby acting as an effective drug. In a fluid system, one can use MD to study the dynamics of the fluid on a nanoscale, helping to elucidate, say, heat and mass transfer [11]. Time stepping algorithms are at the heart of molecular dynamics simulations. Even a modest improvement in time stepping algorithms will result in significant reduction in turnaround time since most MD simulations of real biological systems take days or even months to finish. A closely related technique is called coarsening. Coarsening aims at speeding up simulations by using a coarsened representation of the system, allowing a reasonable compromise between fidelity and speed. Nano-devices refer to systems that have characteristic length of less than 1 micron. Understanding the unconventional physics involved in the operation of such minute devices is critical for the advancement of nano-technology. Unfortunately, fluid flows in such devices cannot be reliably predicted by conventional continuum models such as Navier-Stokes (NS) equations with no-slip boundary condition. Further complicating matters, molecular scale numerical computations of an entire fluid system using an all-atom MD approach are prohibitively expensive. However, an exciting new approach has been proposed recently which uses a hybrid scheme to study the fluids in nano-devices, i.e., the slip boundary layer is resolved using atomic level resolution MD simulations, whereas the bulk of fluid is modeled using the N-S equations [11]. This new approach retains many of the desirable features of the two disparate approaches, MD and continuum models, while overcoming the overwhelming computational expense of a full MD simulation. Nevertheless, the main difficulty with applying this hybrid scheme to study the dynamics of fluids in nano-devices is still the relative inefficiency of the molecular dynamics component. Another exciting application of multiscale MD is in computer–assisted drug design. Medically active drugs typically incorporate small molecules (ligands) that bind to target protein(s) or DNA as tightly as possible. A tight binding leads to a more effective conformational change of the target protein or DNA. This is referred to as binding affinity. For example, in the treatment of breast cancer, the Tomaxifen or Raloxifen molecule binds to the estrogen receptor (ER) tightly so that estrogen molecules can no longer bind to ER, thus preventing cancer cells from proliferating. There may be hundreds or even thousands of candidates for such a drug, however, the binding affinity can only be computed reliably by efficient sampling using very long MD simulations [57]. The investigator plans to continue his research program developing more efficient time stepping algorithms for MD and novel multiscale modeling approaches. These will be applied to two specific problems: modeling the the slip boundary condition in nano-fluid dynamics (for which the proposer