BEDAM/OPLS2005/AGBNP2
Impact-Academic 1.0
Maestro 2014.4
LigPrep 2014.4
Desmond 2014.2
ASyncRE 1.0
UWHAM R-package 1.0
BEDAM Workflow 1.0
Maestro 2016-3
AGBNP2 Implicit solvation model
The work has been performed by the following people:
Rajat Kumar Pal, Emilio Gallicchio; Brooklyn College of CUNY
Lauren Wickstrom; Borough of Manhattan Community College of CUNY
Steven Ramsey, Tom Kurtzman; Lehman College of CUNY
Junchao Xia, Peng He; Temple University
Source of initial structures for BEDAM calculations - The initial
structures for the BEDAM calculations are those deposited in the pose
prediction component of this submission. These initial structures were
obtained by processing them with the Maestro program version 2014
(ProPrep and LigPrep) followed by docking using the Glide program as
detailed in the method description provided with the pose prediction
submission. Complexes were divided into four groups based on the four
receptor structures used for docking.
The ranking predictions were obtained using the Binding Energy
Distribution Analysis Method (BEDAM) with the AGBNP2 implicit solvent
model and the OPLS 2005 force field. System setup was automated using
the BEDAM workflow facility (github.com/ComputationalBiophysicsCollaborative/bedam_workflow)
BEDAM protocol - For the BEDAM simulations, OPLS2005 parameters were
assigned using the System Builder facility of the academic version of
the Desmond program. Structures were energy minimized and thermalized
at 300 K. The complex was defined as any conformation in which the
center of mass the core region of the ligand is within 3.5 Angstroms
of the center of mass of the receptor site. The choice of the atoms of
the ligand and of the receptor that define these centers of mass were
specific to each of the four groups of complexes. Harmonic restraints
with a force constant of 0.6 kcal/mol/Angstrom^2 were applied to the
C-alpha atoms of the receptor to maintain secondary structure. BEDAM
simulations were performed with the Impact program using the replica
exchange molecular dynamics method method in alchemical (lambda)
dimension. We employed 24 replicas at lambda values
0.0,0.002,0.005,0.008,0.01,0.015,0.02,0.0225,0.025,0.03,0.035,0.04,0.07,0.1,0.25,0.35,0.45,0.55,0.65,0.71,0.78,0.85,0.92,
and 1.0) each running at a fixed temperature of 300K. Simulations were
performed one the Computing Grid at Temple University, and the Comet,
Stampede and Supermic XSEDE supercomputing facilities using the
ASyncRE software
(github.com/ComputationalBiophysicsCollaborative/AsyncRE). An average
of approximately 1.2 ns of simulation time per replica were
obtained. Binding energies were collected every 1 ps in,
approximately, the last 1000 ps of production per replica. The binding
energy data was analyzed using UWHAM, which yielded binding free
energy estimates.
Receptor Reorganization Free Energy Offsets - A strong bias with
respect to the initial receptor conformation was observed in the
binding free energy estimates obtained as described above. Reasoning
that the bias was due to a receptor reorganization free energy factor
corresponding to the work of converting the ensemble of the apo
receptor to the given holo ensemble, we conducted limited BEDAM binding
free energy calculations without C-alpha restraints to attempt to
capture some of the effect of receptor reorganization. Four receptor
reorganization free energy offsets, one for each group of complexes,
were applied to the binding free energies obtained with restraints
(see above) by optimizing the degree of linear correlation with the
approximate binding free energies obtained without restraints. The
submitted binding free energy score for each complex is the sum of the
BEDAM binding free energies obtained with C-alpha restraints plus the
reorganization free energy offset assigned to the group the complex
belongs to.
Vanilla Glide
Maestro 2016-3
Glide 7.2
pH 7.4
Tautomers considered
OPLS2005 Forcefield
The content and methodology of this pose prediction submission is
identical (see below for justification) to the ones submitted under
receipt IDs awgis, bckkg, hxm7v and y5gbl by
Steven Ramsey, Tom Kurtzman; Lehman College of CUNY
Emilio Gallicchio; Brooklyn College of CUNY
Lauren Wickstrom; Borough of Manhattan Community College of CUNY
Glide default ligand pose generation
GlideSP scoring function
Maximum number of ligand poses=5
VDW scaling factor=0.8
Glide Default ligand sampling
The BEDAM binding free energy protocol we used
to obtain ligand scores and rankings produces an ensemble of
conformations of the complex, which can not be faithfully represented
by any single structure. Therefore the structures in this pose
prediction submission are the initial structures for the BEDAM
calculations, which are those in the pose predictions submitted under
receipt IDs awgis, bckkg, hxm7v and y5gbl in this challenge.