Research

Crowding effects on protein dynamics

The interior of a cell has a complex environment containing proteins, nucleic acids, lipids and metabolites that occupies ∼ 30% of its volume (concentration ∼400g/L). In this heterogeneous environment the average free space available to molecules is lower than the size of the macromolecules themselves and any reaction that depends on the available volume is affected by this environment. Such an environment clearly differs significantly from the typical protein environment present in an in vitro experimental set up (concentration ∼ 4-16g/L). A question thus arises whether in vivo stability and dynamics of proteins in a crowded medium are very different from that observed in experiments.

Despite having a qualitative understanding of crowding effects on protein folding, several questions remain unanswered: How does the crowding induced influence scale with crowder size and polarity? What are the roles of soft long and short range interactions  between crowder molecules and protein? What is the influence of crowding on the transition state ensemble? We are developing all-atom and coarse grained simulation strategies to answer these key questions.

Markov modeling of protein dynamics

Markov State Models (MSMs) describe protein dynamics in terms of memory less jumps between the states. In this case the continuous MD trajectory is converted into a discreet state space trajectory by a suitable clustering scheme and a transition probability matrix is estimated at a time interval (lag time) that ensures Markovianity of the dynamics. Since the transition probability matrix relies on the local density of conformations, an essential feature of the MSM is that it does not need a long continuous trajectory to predict slow dynamics of a biomolecular system. Instead several relatively short trajectories can be run independently. We focus on improving several aspects of MSM model building and employ the model to study long time scale protein dynamics.

Data driven Langevin modeling

In silico prediction of rare biomolecular events has been a long-standing problem, because slow (say, millisecond timescale) processes are typically out of reach of current all-atom molecular dynamics (MD) simulations. While enhanced approaches may provide an efficient sampling of the system’s free energy landscape, a dynamical description requires some coarse-grained “post-simulation” model that is capable of rebuilding the kinetics from the sampled data. To this end, we employ the data-driven Langevin equation (dLE) approach that has been successfully shown to construct a low-dimensional dynamical model from a given MD trajectory.

References: