Research

Zuckerman Lab research overview

The Zuckerman Lab uses physics-based computational methods to study molecular, meso- and cell-scale systems. Starting from force laws for individual atoms, molecular simulation can provide a detailed picture of nanoscale biological behavior and aid rational drug design. The group develops and applies simulation methods to access important biological timescales and behaviors –such as protein conformational changes associated with ligand binding. The group also uses kinetic models to study "meso-scale" molecular machines, with a focus on the rotary ATPases which remarkably can act as proton pumps or ATP synthases. A new area of interest is modeling complex cellular morphologies and behaviors observed in cancer cells. All the group's work is underpinned by statistical physics and thermodynamics, which govern nano- to micro-scale biology. 

View our recent publications

Parallel simulation across scales

The group is very active in developing and applying weighted ensemble (WE) simulation methods at multiple scales. Ordinary simulations often are hampered by model complexity, which prevents the observation of rare but potentially important events on computable timescales — for example, the folding of a protein or the transition of a cell from a healthy to a pathological state. WE methods, by contrast, orchestrate a parallel set of simulations with periodic pruning and replication interventions that effectively "ratchet" trajectories toward outcomes of interest while simultaneously tracking the unbiased probabilities of such occurrences. The WE approach has already been applied to a wide range of scales, from single-molecule behavior, to virus capsid assembly, to spatially resolved cell-signaling cascades. It is implemented in the WESTPA (The Weighted Ensemble Simulation Toolkit with Parallelization and Analysis) software. Learn more about WESTPA software.

Enhanced Parallel Simulation for Molecules, Cells, and Networks
The "weighted ensemble" sampling approach makes possible computations of long timescales and mechanisms of rare events which otherwise would not be feasible. It applies to simulation problems across many scales in biology.

Induced-fit docking

Mixed-Resolution Modeling: all-atom where it matters, coarse-grained elsewhere. Example shown: Ligand-binding domain of the estrogen receptor alpha
Mixed-resolution modeling is a way to "right size" a model to the computational problem at hand — for example by mixing coarse and atomistic descriptions for a drug target, enabling fast and relatively accurate ligand docking.

The interactions of small molecules with protein receptors are crucial for cell physiology, as well as for drug design, but flexible or poorly resolved protein structures may not be amenable to standard computations. The group has developed a mixed-resolution software platform that avoids the artificial rigidity of standard docking protocols as well as the computational expense of fully detailed molecular dynamics. The binding site of the protein is represented with full atomistic detail and flexibility to capture ligand interactions faithfully, but the remainder of the protein is simplified in a coarse-grained framework.  

Kinetic modeling of molecular motors: The rotary ATPases

Nature has evolved remarkable molecular machines for synthesizing ATP and, when functioning in reverse, for pumping protons or ions. These "meso-scale" complexes consist of 20 or more proteins, but can be studied with relatively simple discrete-state kinetics models. What are the advantages of these large complex machines compared to simpler alternatives? What biophysical mechanisms can account for the variable proton:ATP stoichiometry observed experimentally? The group is addressing these and other questions.

Kinetic Modeling
Molecular machines perform an incredible array of functions within the cell, but typically are too large to model computationally in atomistic detail. Discrete-state models, such as for ATP synthesis by the ATP synthase, are an effective means to study mechanistic hypotheses for such machines.

Non-equilibrium statistical mechanics of trajectories

Underlying the WE method discussed above is the fundamental physics of trajectories. Harnessing theory is critical not only for advancing sampling methods like WE, but also for analyzing trajectories. It is increasingly common to generate a number of short molecular simulation trajectories and "stitch" them together using approximation methods to infer long-timescale behavior. The group has recently pioneered non-Markovian trajectory analysis that permits much more accurate inferences of kinetic and mechanistic behavior than had previously been possible, and we are working to apply the methods more broadly and make them readily available.