Hierachical Natural Move Monte Carlo using MOSAICS

After having recently published a large scale Molecular Dynamics simulations project of TCRpMHC [1,2] interaction I have extended my research to another technique of spatial sampling. At this week’s group meeting I presented the first results of my first MOSAICS [3] project.

The MOSAICS package is a software that allows for so called hierarchical natural Monte Carlo moves. That means that the user can specify regions in the protein of interest. These regions are indented to reflect “natural” sets of atoms and are expected to move together. An example would be a stable alpha-helix. “Hierarchical” means that region can be grouped together to super-regions. For example a helix that is broken by a kink [4] in its middle could have a region for the helix parts on both sides of the kink as well as for the overall helix. An example for peptide/MHC is illustrated below.


MOSAICS uses Monte Carlo moves to rearrange these region with respect to each other. A stochastic chain closure algorithm ensures that no chain breaks occur. An example of such movements in comparison to classical all-atom Molecular Dynamics is shown below.


In this study we used MOSAICS to simulate the detachment of peptides from MHCs for experimentally known binder and non-binder. An example of such a detaching peptide is shown below


Our results show that experimentally known non-binding peptides detach significantly faster from MHC than experimentally known binding peptides (results to be reported soon).

As a first conclusion of this project:
After having worked with both MOSAICS and Molecular Dynamics simulations, I think that both techniques have their advantages and disadvantages. They are summarized below:


Which technique should be chosen for which project depends mainly on what the aims of these projects are. If large moves of well defined segments are expected then MOSAICS might be the method of choice. If the aim is to investigate fine changes and detailed dynamics Molecular Dynamics simulations might be the better choice.


1.    Knapp B, Demharter S, Esmaielbeiki R, Deane CM (2015) Current Status and Future Challenges in T-cell receptor / peptide / MHC Molecular Dynamics Simulations. Brief Bioinform accepted.
2.    Knapp B, Dunbar J, Deane CM (2014) Large Scale Characterization of the LC13 TCR and HLA-B8 Structural Landscape in Reaction to 172 Altered Peptide Ligands: A Molecular Dynamics Simulation Study. PLoS Comput Biol 10: e1003748.
3.    Sim AY, Levitt M, Minary P (2012) Modeling and design by hierarchical natural moves. Proc Natl Acad Sci U S A 109: 2890-2895.
4.    Wilman HR, Ebejer JP, Shi J, Deane CM, Knapp B (2014) Crowdsourcing yields a new standard for kinks in protein helices. J Chem Inf Model 54: 2585-2593.

Leave a Reply