In the near future, the Schulten group will carry out simulations of the pump cycle of bacteriorhodopsin in the context of the complete bacterial (purple) membrane. Schulten seeks to model over the next several years the entire photosynthetic unit of purple bacteria including, besides a membrane-water system, multiple proteins (photosynthetic reaction center, light harvesting systems I and II, bc1 complex, ATP synthase, cytochrome), jointly turning light energy into ATP [6,48]. Individual components of these systems were studied in the Schulten group during the last decade, i.e., the lipid bilayer [49,50], the photosynthetic reaction center [3], the light harvesting complexes [51,42,51], the bc1 complex [8], but placing proteins in their integral setting will shed light on essential processes in cells, like the self-aggregation of cellular structures discussed in [52], the cooperation of bioenergetic proteins and the use of chemiosmotic potentials.

Multi-protein systems and other biomolecular aggregates of particularly large size have been resolved structurally, for example, complete virus capsids, ATP synthase, multienzyme systems of the citric acid cycle, and protein folding chaperones like GroEL/GroES; most likely the atomic level structure of the ribosome will soon be solved. These exciting and important structures involve 10 5 -106 atoms and are out of the reach of most modeling programs. Schulten is a pioneer who has harnessed the power of modern parallel computers to model extremely large structures and one can easily predict that his group will play a central role in computational biology in this regard during the next decade. In doing so the group will benefit from the best environment for scientific computing in the US, the U. of Illinois at Urbana-Champaign.

Schulten explores multiresolution descriptions of large biopolymer systems. He and his coworkers have developed a so-called vector quantization method for topologically faithful discrete representations of high dimensional manifolds [18] and have applied it to biopolymers [53]. Schulten's group has also developed recently a computationally inexpensive method to describe DNA strands as charged elastic rods and succeeded in modeling the lac repressor - DNA complex with a ~ 90bp DNA loop [11]; the group will soon model nucleosomal DNA with this method.

The self-aggregation of biomolecular aggregates in cells, the mutual recognition of biopolymers, and the adhesive forces involved are essential attributes of living systems that pose deep conceptual challenges and are of great medical relevance. Schulten wants to approach the related problems by mechanically manipulating complexes of biopolymers in simulations. During the past three years his group has developed and applied so-called steered molecular dynamics (SMD) simulations that add user-defined forces to simulations and can probe mechanical properties of proteins and adhesive interactions [54]. SMD calculations have already provided important qualitative insights into biologically relevant problems for applications ranging from the identification of ligand binding pathways [55 ,56,57,58,59] to the explanation of elastic properties of proteins [13,60,61,62,47]. SMD, in particular, revealed the participation of amino acid side groups in guiding biotin stepwise into its avidin binding site [55], discovered the binding path of retinal in bacteriorhodopsin from the lipid phase, rather than the aqueous phase as erroneously believed for decades [56], suggested that ATP hydrolysis in actin [57] as well as hormone binding in certain nuclear hormone receptors [58] proceed by a back-door mechanism, and has investigated the motion of a key (Rieske iron-sulfur protein) domain of the bc1 complex that redirects electron transfer reactions between three redox centers [8].

Recently, Schulten has taken SMD an essential step further. His group has succeeded in carrying out SMD manipulations interactively by combining the group's molecular graphics program VMD and molecular dynamics program NAMD2, permitting simulation of a graphically represented system during viewing. In this new approach to biomolecular modeling a user can define forces through a six degree of freedom pointer, a so-called haptic device, that senses the forces resisting the user's manipulation of the actual simulation. The method is presently applied to study selective conduction in ion channels, in which case a user pulls various ions through channels defined through several thousand atoms being simulated during a session on powerful parallel machines. Another ongoing study involves the docking of a substrate to a protein, an application that should lead to a crucial tool for drug design. Schulten believes that such interactive modeling also furnishes the adequate tool to investigate structurally resolved nanometric biomolecular machines, i.e., their mechanical properties and adhesive interfaces. Interactive modeling is also an extremely timely tool to complement experimental methods that manipulate single biopolymers, e.g., atomic force microscopy and optical tweezers. Several ongoing projects in the Schulten group are directly related to such observations. Schulten expects his new method, through its natural graphical/haptic user interface and intuitive appeal, will furnish a tool that will be as widely used by researchers in biomedicine tomorrow as molecular graphics is today.