Reverse transformation
- Details
- Last Updated: Thursday, 22 August 2013 08:59
Reverse transformation
In this module we explore the use of the reverse-transformation tool, to convert between coarse-grained and fine-grained representations of a system.
Simulations at coarse-grained level allow the exploration of longer time-scales and larger length-scales than at the usual atomistic level. However, the loss of detail can seriously limit the questions that can be asked to the system. Methods that re-introduce atomic details in a CG structure are therefore of considerable interest. Such structures provide starting points in phase space for simulations at the more detailed level that may otherwise take too long to reach. As a demonstration of such a method, we will use restrained simulated annealing (SA) to increase the resolution of a system containing a WALP peptide spanning a DPPC bilayer. We will investigate how the number of steps allowed for the reconstruction influences the quality of the generated FG structure.
Reverse transformation: transformation run
In this part of exercise we will use a modi ed version of gromacs that allows one to generate a FG structure from CG beads*. The same version is also used to transform a FG structure to CG beads, section 1e. Thus, the program allows one to switch between FG and CG representations. To achieve this, additional information is put in the topology le at the FG level in a section called [ mapping ]. The tool pdb2gmx can generate this mapping for proteins automatically. Note that this will usually require the option -missing to be speci ed! Two new functions in mdrun are responsible for restrained simulated annealing (SA) to converge from a random FG structure to a FG structure compatible with the CG structure, i.e. the FG positions are optimized in such a way that the CG structure is preserved in the FG-to-CG mapping (each CG bead is at the center of mass of the FG atoms that map to it). There is also small tool, called g_fg2cg, to generate CG structures from FG ones defi ned by the mapping and random FG starting structures based on CG input le, that are afterwards optimized during a SA run.
The source code for this version of the script can be found Downloads/Tools section of this website.
The files needed for this section can be downloaded from:
Transformation run
1. Unpack the rev_trans.tar.gz in the MARTINITUTORIAL directory that contains all necessary gromacs files for this exercise.
2. Compile and/or source the modifi ed version of gromacs (remember this tool is based upon gromacs version 3.3.1 and needs the corresponding tricks and threats to be compiled.)
source /where-ever-you-installed-it/gromacs-3.3.1/bin/GMXRC
export GMXLIB=/where-ever-you-installed-it/gromacs-3.3.1/share/gromacs/top
3. Modify the FG fg.top fi le in such a way that the number of water and lipid molecules is the same as in the coarse-grained model. One FG_W corresponds to four normal water molecules.
4. Use g_fg2cg to construct an input atomistic structure for a simulated annealing run. The coarse grain structure is already prepared for you and is called cg.gro. Check the output with ngmx or VMD.
g_fg2cg -pfg fg.top -pcg dppc_cg.top -n 0 -c cg.gro -o fg.gro
5. Use grompp to create a topol.tpr file.
6. Perform a SA run by typing
mdrun -coarse cg.gro -v
7. Change the number of simulation steps to 1000 and SA time parameters to 0 1.5 0 1.5 0 1.5 in your mdp file. (see comments for additional mdp options at the bottom of fg.mdp file) Perform another SA run with the altered parameters.
8. Change the number of simulation steps to 5000 and SA time parameters to 0 7.5 0 7.5 0 7.5. Perform a SA run again.
9. If you have enough time, change the number of simulation steps to 60000 and SA time parameters to 0 100 0 100 0 100. Repeat the SA run.
10. Plot the potential energies for all runs and compare them.
11. Compare FG WALP models from all runs. Check secondary structure and phi and psi angles.
12. Compare dihedral distribution for one of dihedral angles from the tail of the DOPC lipids.
*A.J. Rzepiela, L.V. Schäafer, N. Goga, H.J. Risselada, A.H. de Vries, S.J. Marrink, Reconstruction of atomistic detail from coarse grained structures, J. Comput. Chem., 31, 1333-1343 (2010)