Sampling functional motions of proteins
Title: Sampling functional motions of proteins
SNIC Project: SNIC 2014/1-30
Project Type: SNAC Medium
Principal Investigator: Berk Hess <>
Affiliation: KTH Royal Institute of Technology
Duration: 2014-02-01 – 2015-02-01
Classification: 10402 20301


One of the main application areas of molecular simulation is the understanding of functional motions in proteins. Molecular dynamics can reveal molecular motion in atomistic detail, but this comes at a high computational cost. Billions of time steps of femtoseconds are required to reach biologically relevant time scales of microseconds. For particular scientific problems smarter sampling could accelerate sampling by orders of magnitude, thereby enabling new problems to be tackled. We have adapted a weight-histogram method to work on sampling of conformational changes in bio-molecules and extended the method to work on multiple replicas of a system simultaneously. Given one or more generalized degrees of freedom, the method samples those and produces free energy profiles. The method has been shown to work on a model system of peptide folding (manuscript in progress). Our target applications are slow functional motions in proteins, which are of high scientific and pharmaceutical interest and where large gains can be made. We have just started applications to the ligand gated ion channel GLIC and the calcium ATPase SERCA. For GLIC we are interested in how mutations affect the free-energy profile for passing of ions. Experiments can measure the transport, but they can't explain it. For SERCA we want ot find out where and how calcium ions enter the protein. This is currently unknown. Initial runs with our method on SERCA have shown that calcium can enter through a hypothesized pathway which is initially completely closed. There is another potential pathway which is hydrated. We need longer simulations to determine which pathway is more probable and to resolve all dtails of the pathway. A second project is studying the atomistic aspects of wetting of surfaces. The past year we have studied the wetting of substrates with different atomistic structure. We used a quasi-2D setup of water droplets of 100 nm radius (3.5 million atoms). This large size is required to avoid finite size effects. We observed that with the same equilibrium physics, we can get different dynamics. A paper for Phys. Rev. Letters is in preparation. We are contiuing this work and also moving to electrowetting. Electric fields can modulate the wetting. On continuum scales this is well understood. But most issues with application occur when the continuum description breaks down. To understand what happens, the structure and transport of ions dissolved in the liquid needs to be studied, as ions interact strongest with the electric fields. Molecular simulation is the suited tool for this. Here we will initially need somewhat smaller systems with about a million atoms. Finally, I would like to use some of the time of this allocation for testing the extreme parallelization of the Gromacs code, of which I am one of the main developers. Triolith is the ideal platform to test what a state-of-the art hardware+code is capable of. We published a paper on Gromacs 4.5, submitted a paper for version 4.6 and will work on a paper on parallelization limits.