SDA
Simulation of Diffusional Association
(version 6)

SDA example run "aubs", "aubs-pqr"

This example run can be found in directory examples/aubs and examples/aubs-pqr
The following subdirectories are present in these directories:

Barstar - Au(111) surface docking example

2 types of simulations are done for docking barstar to Au(111) surface and given in the ../../aubs and ../../aubs-pqr directories:

1) the electrostatic grid is computed with UHBD using standard partial charges from the qtable in the ../../aubs subdirectory;
../../aubs/scr/aubs-prepare-and-run

2) the electrostatic grid is computed with UHBD using partial charges from the pqr file (p2.pqr) for barstar at pH=7.4 generated with the web-based H++ program (http://biophysics.cs.vt.edu/H++/):
../../aubs-pqr/scr/aubs-prepare-and-run

The script also contains the commands (commented out) for calculating the electrostatic grid with APBS instead of UHBD.

The results of the SDA run can be found in the run-hits/ subdirectory or can be generated in a new run/ subdirectory using the script ../scr/aubs-prepare-and-run .  In order to execute this script locally, you need to define 2 parameters - the location of the SDA6 distribution directory and the location of the UHBD/APBS executable.

To get an overview of the protein-surface docking properties (conformations, energies etc.) you may like to group docking complexes into clusters.
For this, all structures stored in the file "complexes" must be shifted to the center of the surface plane: x=0,y=0 using the python script DelRotZ_complexes.py (this step is not needed for the protein-protein docking!). The output file wtrotz.complexes has the same format as an original file complexes.

Then there are two ways to proceed with clustering using the "wtrotz.complexes" file (or "complexes" for protein-protein binding):
I) the faster way is to use python script that allows to cluster all complexes with RMSD=5 Angstrom :myClustering_complexes.py (to change RMSD you have to change the RMSD_CUTOFF parameter in the python script myClustering_complexes.py). The output file "mycluster.complexes" can be then used to generate pdb files of the representative structures from each cluster with program generateFortComplexesPdb.py.
II) another possibility is to use program ../../../bin/clust that allows group complexes into several clusters. The number of clusters required is the input parameter. An example of using the clust program is included in the script ../../aubs-pqr/scr/aubs-prepare-and-run (note, that the input file should be "wtrotz.complexes" for the protein-surface case and "complexes" otherwise).
The results of two clustering procedures applied to the output of the example scripts are compared below.

Clustering results generated by the python scripts myClustering_complexes.py for the case "aubs" and "aubs-pqr".

The contributions of different energy terms to the barstar-Au binding energy for the first three binding configurations can be found in the output file mycluster.complexes :

1) Run with electrostatics based on the partial charges from the qtable ("aubs" example, BS net charge is -5.88)

Cluster	Lowest binding energy in the cluster /kT	LJ energy /kT	Electrostatic energy /kT	Surface desolvation energy /kT	Average binding energy in the cluster /kT	Cluster population
I	-0.5087E+02	-0.1103E+03	-0.6251E+01	0.6982E+02	-45.092	611
II	-0.5066E+02	-0.7335E+02	-0.3870E+02	0.6426E+02	-42.308	105
III	-0.4131E+02	-0.7703E+02	-0.1910E+02	0.5765E+02	-37.685	1246

2) Run with electrostatics based on the partial charges from the pqr file computed with H++ at pH=7.4 ("aubs-pqr" example, BS net charge is -6)

Cluster	Lowest binding energy in the cluster	LJ energy	Electrostatic energy	Surface desolvation energy	Average binding energy in the cluster	Cluster population
I	-0.5036E+02	-0.1096E+03	-0.5818E+01	0.6908E+02	-44.349	711
II	-0.4874E+02	-0.7404E+02	-0.3606E+02	0.6418E+02	-42.306	648
III	-0.4195E+02	-0.7755E+02	-0.1924E+02	0.5763E+02	-39.165	600

Although the binding energies are slightly different in the "aubs" and "aubs-pqr" examples, binding configurations are practically the same (see fig. below, structures of "aubs" and "aubs-pqr" are shown in cyan and pink, respectively).

The first binding configurations is mainly driven by the LJ term, while for the second one the electrostatic term is more important.

Binding configuration of the cluster I

Binding configuration of the cluster II

Binding configuration of the cluster III

Clustering results generated by the program "clust" for the case "aubs" .
Note, that the program clust orders clusters by their populations, while the python script orders the clusters by total energy. Thus, the clusters No 2,3, and 1 generated by the clust program (see table below) correspond the clusters No 1,2, and 3 generated by the python clustering scripts, respectively (see table above). The further differences of two clustering program are summarized in the table below

Output information of the clust program:

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

No	ClSize	ClFSize	Repr	ReprE	ClAE	CLAED	RepRMSD	CLFRMSD	ElDesE	HyDesE	LjE	CouElE	SuDesE	tElE
1	1729	388973	1791	-35.930	-33.076	6.907	30.189	30.182	11.340	-2.724	-76.44	-11.150	60.110	-16.880
2	507	345155	216	-46.980	-46.223	13.786	31.455	31.523	16.580	-4.054	-108.800	-8.538	71.790	-5.906
3	214	52671	580	-41.610	-34.567	2.99	30.679	30.342	19.320	-2.743	-74.730	-12.640	68.500	-32.640

	python scripts	clust program
output file	mycluster.complexes	standard output
clustering parameter	RMSD (5 Angstrom default)	the number of the output clusters
cluster ordering	low-energy first	high populated first
energy terms in the output file	those of the complex with the lowest total energy	those of the structural representative complex with smallest RMSD relative to all members of the cluster

January 2010