Data from molecular dynamics simulations for COSMOmic

Introduction

COSMOmic is an extension of COSMO-RS for anisotropic systems such as membranes and micelles, it was introduced by Klamt et al. and detailed information can be found in their original article [DOI: 10.1021/jp801736k ]. The COSMOmic method is implemented in the COSMOtherm program of COSMOlogic.

COSMOmic requires as input the structural composition of the anisotropic system. More precisely, the atomic distribution for each atom type of the system is needed, usually calculated from a snapshot of a molecular dynamics (MD) simulation. COSMOtherm takes as input the coordinates of the atoms as pdb file (single snapshot). However, we have shown that using averaged (over a MD trajectory) atomic distributions are more reliable and scientifically more reasonable [DOI: 10.1002/jcc.23262]. Whereas some groups provide pdb files of lipid membranes from MD simulations, atomic distributions suitable for COSMOtherm or trajectories from which the distributions could be calculated can usually not be found. Therefore, here we provide depth-dependent membrane and micelle compositions averaged over MD simulations carried out in our group. In the following, we assume that you know our article which discusses the connection of MD simulations and COSMOmic:

S. Jakobtorweihen, T. Ingram, and I. Smirnova,
Combination of COSMOmic and molecular dynamics simulations for the calculation of membrane-water partition coefficients,
J. Comput. Chem. 2013, 34, 1332-1340.

Provided data

Below you will find a table that contains data and information on different lipid bilayer systems. A snapshot is shown in the first column. Information about the system and the MD simulation can be found in the second column, the simulation time refers to the time over which the atomic distribution was averaged. If a salt concentration is given, it refers to the concentration in the water phase (outer layers), and not the concentration in the whole system. Averaging was started only after an appropriate equilibration time. Three different files are provided for each system: (1) the averaged atomic distribution in COSMOtherm mic format, (2) a single system snapshot in pdb format, and (3) a lipid/surfactant conformer taken from the MD simulation. The latter is provided, as COSMO-RS needs at least one conformer for each species in the system and it has been shown that conformers of amphiphilic molecules should be taken out of the correct environment (see DOI: 10.1002/jcc.23262 and references therein). Therefore, the lipid conformer was chosen from the lipid bilayer MD trajectory as described in our article [DOI: 10.1002/jcc.23262]. Analog, the surfactant conformers were taken from a micelle simulation. The lipid/surfactant conformer can be used to calculate a cosmo file to be used in the COSMO-RS calculations.

Usage in COSMOtherm

Below you find mic files (COSMOtherm 14 and later) each containing an averaged atomic distribution. Before you can use these files, you have to adjust the list of cosmo files. If you are using COSMOmic via COSMOthermX, you can load the mic files with Read Micelle. If you are using the command line, the mic file name has to be given in the input file (RMIC=).

The data in the mic files is organized as follows: The first line defines variables needed to characterize the bilayer or micelle, see the COSMOmic manual for further details. Note that the temperature given in this line is the temperature the distribution was created at. The temperature of the COSMOmic calculations is specified elsewhere (inp file). On the second line the list of cosmo files starts. For each compound making up the system a cosmo file is needed. Here you have to add the correct name and location of your cosmo files. It is important that you do not change the order of the file list, as it has to correspond to the atomic distributions. Nr lines follow this list of cosmo files, where nr is the number of layers. That is, each line defines one layer, starting from the membrane/micelle center to the water phase. Each line has as many columns as atom types in the bilayer or micelle. For example for a DMPC/water system you will find 121 columns, as DMPC has 118 atoms and water has 3 atoms. In all provided files the first columns are the lipid or surfactant atoms and the last three columns are the water atoms (hence, the third last column represents the oxygen atoms of water). The order of the lipid/surfactant atoms can be seen in the conformer file. Each column sums up to one, as the values are the fractions of the atom types that are found in a specific layer. The lipid/surfactant conformer used in the calculation has to fit to the distributions; the order of the atoms should be the same. If you use your own conformer with the provided data, you have to take care that the atom order is correct. The provided conformer has the correct order. However, you have to generate a cosmo file (e.g., single-point Turbomole calculation) to use it for a COSMO-RS calculation. Please note, the pdb files contain the atom names as used in the MD force field; in addition the element symbol is included in the conformer files. Sometimes older versions of TmoleX have problems with these atom names (ERROR: No radius for … found). If you encounter this problem, you might wane use Open Babel to convert the pdb file to a xyz file and use that as input.

References

If you use averaged atomic distributions or material from this web page, please cite:

• S. Jakobtorweihen, T. Ingram, and I. Smirnova,
  Combination of COSMOmic and molecular dynamics simulations for the calculation of membrane-water partition coefficients,
  J. Comput. Chem. 2013, 34, 1332-1340.


• If the material was first used in another publication, you will find references in the last column of the table.


• A Klamt, U. Huniar, S. Spycher, and J. Keldenich introduced the COSMOmic method, the original article you can find here. The COSMOtherm program you find here. An extension of COSMOmic to ionic solutes is also interesting [DOI: 10.1021/jp509348a ]. By default the here provided mic files do not use the electrostatic potential (EPOT parameter not included).

Micelles

For micelles, instead of the simulation time, the number of configurations N_conf over which the atomic distribution was averaged is shown. The reason is that micelles sizes can fluctuate over the simulation and the averaging is done over all micelles of the same size found in a simulation [DOI: 10.1016/j.fluid.2016.03.006]. The specific micelle size provided here is chosen according to the criteria discussed in our article [DOI: 10.1016/j.fluid.2016.03.006], where a spherical shape is one criterion. Hence, it is not necessarily the experimental aggregation number. The averaged eccentricity ε as a measure for the spherical shape of the micelle is also provided in the table. Furthermore, the table shows the aggregation number of the micelles N_agg for which the atomic distribution was calculated. In addition, the number of surfactants in the simulation N_surf is shown. In some cases the number of surfactants in the system is higher than the aggregation number (N_agg ≠ N_surf), see also [DOI: 10.1016/j.fluid.2016.03.006]. Hence, the system snapshot contains N_surf surfactants and the atomic distribution is calculated for micelles with N_agg.

If you use averaged atomic distributions for micelles or material from the table below, please consider this article:

• E. Ritter, D. Yordanova, T. Gerlach, I. Smirnova, S. Jakobtorweihen,
  Molecular dynamics simulations of various micelles to predict micelle water partition equilibria with COSMOmic: Influence of micelle size and structure,
  Fluid Phase Equilib. 2016, 422, 43–55.

Force field parameters for Triton were introduced in this article:

• D. Yordanova, I. Smirnova, S. Jakobtorweihen,
  Molecular Modeling of Triton X Micelles: Force Field Parameters, Self-Assembly and Partition Equilibria,
  J. Chem. Theory Comput. 2015, 11, 2329–2340.