Molecular MARTINI

Molecular simulations can be computationally expensive. As we increase their length and time scales, running simulations becomes increasingly slow or requires increasingly more resources. Improved software and hardware reduce the cost, but these are not the only ways. Another way is to use a less expensive model. The Martini force field is such a model, and I have been involved in some of its recent developments.

In most models used for classical molecular simulation, molecular systems are represented with interaction sites corresponding to the individual atoms. Martini is a coarse-grained force field; each interaction site represents a group of atoms. With fewer interaction sites to compute and a smoother free energy surface, simulations using the Martini model can be orders of magnitude faster than the same simulations using a model that represents the atoms individually.

When we resolve the structure of molecules, we find where the individual atoms are located relative to each other. We do not know the structures in the Martini representation. Instead, we need to build them from atomic coordinates. I contributed to Martinize 2 and Vermouth, tools to transform topologies from one model to another. The tools now have their paper in eLife.

Martini is often used to simulate lipid membranes, of which cholesterol is an essential component. However, version 3 of the Martini force field introduced a regression in the treatment of cholesterol, leading to unrealistic results. I was part of an international collaboration to fix this issue. The result is now published in the Journal of Chemical Theory and Computation.