aubs example

This example shows the implementation of the ProMetCS force-field in SDA 7.
It is parameterized for protein-metal surface interaction, including electrostatic image charge and metal desolvation energy terms. The ProMetCS force field is described in : Kokh et al.ProMetCS: An Atomistic Force Field for Modeling Protein−Metal Surface Interactions in a Continuum Aqueous Solvent.J. Chem. Theory Comput., (2010) 6:1753-1768.i DOI: 10.1021/ct100086j

It has been implemented in SDA 7 and is available for multiple proteins simulations (SDAMM). It is still under development.
We advise to compare your results with the SDA 6 released version for reliable results.

This example demonstrates calculations for (i) the diffusional docking of barstar (BS) to an Au(111) surfacei and (ii) the simulation of the diffusion of multiple BS molecules in the presence of an Au(111) surface.

Directory description

This example can be found in directory examples/aubs.

prepare_grids_and_ecm/	Directory for generating grids and effective charges needed for running the simulation
data_grid/	Data files (pdb and input files) needed for running the examples
run_hits/	Results of the example run script for protein-metal surface docking
run/	Directory to run the protein-metal surface docking
run_sdamm_hits/	Results of the example run script for a multiple protein simulation
run_sdamm/	Directory to run the multiple protein simulation
doc/	This documentation
unit-test/	For developers, regression tests for many different combinations of interactions. The grids need be generated first.

Detail of the input files

The different components of the ProMetCS force field have been implemented in SDA 7 with the possibility to be selected individually:

Electrostatic interaction with a metal surface: the electrostatic image-charge method is applied. It is a term which can be applied with any conductor surface.
It must be activated for each type of solutes in the sda input file (image_charge = 1).
At short distance an analytic correction, modifying the dielectric constant close to the surface, can be activated globally, for all solutes (correction_image_charge = 1), at the exception of the surface.
The metal desolvation term, accounting for the desorption of the water molecules, can be adjusted with the input parameters in the group MetalDesolvation of the input file. It applies to all solutes, except the surface.
The electrostatic desolvation image-charge is implemented in a similar way to the electrostatic image-charge, but with the electrostatic desolvation grid and the square of the effective charges.
It is activated by a global input parameter (desolv_image_charge = 1) and applies to all solutes except the surface.

Running examples

Assure you have compiled all the executables and tools in sda_flex/bin/ first or refer to the compilation section.

Run first the script run-aubs_ed_hd_ecm_lj.sh from the prepare_grids_and_ecm directory to generate the grids necessary for this example
Note : the generation of the Lennard-Jones grids need more time than the other grids (approx. 1-2 hours)

Then go in one of the directory containing input file (run or run_sdamm) and run the shell script

For the protein-surface simulation :


cd run/ 

./run_sda_prometcs.sh

and for the multiple proteins-surface simulation :


cd run_sdamm/

./run_sdamm_prometcs.sh

You can check that output files are similar to the ones generated in the *_hits directory.

Analysis of simulation results for 2 solutes

For visualization and analysis of the docking configurations of BS on the gold surface, a clustering procedure can be used.
For this one has to eliminate rotational (around Z) and translation (in (X,Y) plane) degrees of freedom of BS, since the interaction with the gold surface depends practically only on the Z coordinate (scripts DelRotZ_complexes-SDA7.py).
Then one can use either a k-means clustering (as in the example case bnbs) or a hierarchical clustering procedure (python code myClustering_complexes-SDA7.py, see script run_sda_prometcs.sh). The latter method enables clustering of the docking orientations with a pre-defined RMSD (5 Angsrom, for C_alpha) within each cluster.

In this example, 24 clusters were identified from 2000 low-energy docking positions (see out_surface_clustering). The clusters are ranked by binding energy; the first four clusters have the lowest average binding energy and high population:

(cluster #)	(average energy)	(population)
1	-4.149e+01	2.419490e+05
2	-3.030e+01	7.127600e+04
3	-3.880e+01	7.419120e+05
4	-3.219e+01	8.825750e+05

(Note that using a smaller number of low-energy complexes for clustering (for example, only 500), leads to fewer clusters (6 in the case of the 500 complexes analyzed). Nevertheless, the first 3-4 low-energy clusters will be preserved, though the average energy, population and order may be slightly different. (These clusters also correspond to the first three low energy clusters obtained in similar simulations with the SDA6).

The most populated cluster #3 corresponds to the docking position with the strongest van der Waals interaction between BS and gold (in particular, such strong binding residues such as TRP44 and TYR47 are involved) and the weakest electrostatic interactions;

Clusters # 1 and # 2 have similar BS docking configurations (only slightly differently tilted) that are mediated by electrostatic interactions (due to charged residues that go inside the gold hydration shell which is characterized by a smaller dielectric constant than that in the bulk water). In this example, up to three charged residues are involved in binding: ASP35+ASP39+GLU80 for cluster #1 and GLU80+ASP39 for cluster #2. Usually, the docking configurations arising from such an electrostatically-driven anchoring of a protein by several charged residues have quite a narrow energy funnel and are less populated than protein docking positions driven by the LJ binding term;

Cluster #4 represents an intermediate case with higher energy (see below) in which binding is mediated by ASP35,ASP39,GLU80 and TYR29;

The magnitudes of the energy terms for the low-energy representatives of each cluster can be found in mycluster.comp_cluster) :

(cluster #)	(total energy)	(Lennard-Jones + Metal desolvation; kT)	(Image-charge + electrostatic desolvation; kT)
1	-44.0	-3.9	-37
2	-42.5	-5.8	-35
3	-42.2	-36.7	-1
4	-34.6	-15	-17

The docking positions of BS on gold for each cluster were generated by trajectory2dcd:

cluster #1 & #2	cluster #3	cluster #4

Protein residues within 5.5 Angstrom of the gold surface are shown in stick representation.

Results of multiple protein simulations

The extension of SDA 7 to multiple protein simulations in the presence of a gold surface is currently under development. The example demonstrates how simulations of this system can be run but the correct parameterization for protein-gold and protein-protein interactions cannot currently be specified in the sda input.
This type of simulation allows different arrangments of the solutes with the surface to be studied at a high protein concentration.

Snapshot after 100 ns of simulation. 2 solute barstar molecules are bound to the surface in configurations similar to clusters #2 and #7 from the simulations with one barstar molecule

[Back to Main Documentation]

Privacy Imprint