FAAH from WCG: “FightAIDS@Home – Phase II”

New WCG Logo


No writer credit found


FightAIDS@Home Phase II is a milestone in the collaboration between Dr. Arthur Olson and the Molecular Graphics Laboratory at the Scripps Research Institute and Dr. Ronald M. Levy’s group at Temple University.


Olson Lab

Levy Group at Temple

Dr. Olson initiated the largest HIV virtual screening effort ever using his computational molecular docking software, AutoDock, over a decade ago using IBM’s distributed volunteer computing grid, the World Community Grid (WCG), called FightAIDS@Home.


As director of the HIV Interaction and Viral Evolution (HIVE) Center, an NIH funded HIV collaborative research center, the virtual screening results play a significant role informing a portion of the Center’s research.


The HIVE Center comprises a multidisciplinary team of scientists whom together aim to elucidate the structural and dynamic relationships between interacting HIV macromolecules in an effort to design future therapeutics to combat HIV and its evolution of drug resistance. Dr. Ronald Levy brings to the HIVE Center over 30 years of leadership and experience in the development and application of molecular dynamics simulations to study the structure and dynamics of proteins and their complexes.

During FightAIDS@Home Phase I, millions of compounds have been screened against HIV related protein targets, and thousands of potential drug candidate compounds have been identified. In Phase II, we plan to refine the selection of these thousands of possible compounds by performing more detailed computational experiments using our molecular dynamics engine IMPACT, and the Binding Energy Distribution Analysis Method (BEDAM).


The introduction of BEDAM simulations into FightAIDS@Home Phase II presents an enormous opportunity to refine and enrich the results from Phase I, but also presents a technical challenge as the constraints on the simulations running on the WCG are much different than the constraints on the simulations when running on more conventional computational resources.

The first experiments of FightAIDS@Home Phase 2 seek to achieve two goals: first, to confirm that the new simulation schema is working as intended and gives sufficiently reliable results compared to traditionally run simulations; second, to demonstrate that using BEDAM in conjunction with AutoDock results in better predictions than using AutoDock or BEDAM alone. There exists a symbiotic relationship between docking and more computationally demanding free-energy methods like BEDAM—without docking, the computing time required to score thousands of ligands with free energy methods is intractable, and without free energy methods, the relationship between docking scores and experimental binding affinities remains much more empirical and less accurate.

Overview of Molecular Dynamics and BEDAM
Physical free energy models of binding

One key aspect of drug discovery is to identify compounds which bind strongly and specifically to the target receptor, and therefore there exists considerable interest in the development of computational models to predict the strength of protein-ligand interactions. Thermodynamically, this interaction strength between ligand molecule and receptor is measured by the binding free energy, and many computer models aim to predict the protein-ligand binding free energy by simulating the interactions between protein and ligand in a bound complex. Docking methods use empirical scoring functions to estimate binding free energies in order to distinguish between between ligands that bind strongly from ligands that bind weakly or not at all.

On the other hand, physical free energy models, which use physics based effective potentials, seek to compute accurate protein-ligand binding free energies based on the principles of statistical mechanics. Unlike docking methodologies, many of these methods are exceptionally computationally intensive and are highly dependent on both accurate modeling of interaction force fields (ligand-ligand, ligand-solvent, protein-protein, protein-solvent, and protein-ligand) and efficient sampling of all rotational and translational, internal and external, degrees of freedom of the ligand and protein.

The statistical mechanics theory of binding provides a prescription to compute the binding free energy from first principles, which can be implemented in various ways. One method developed in the Levy lab is called the binding energy distribution analysis method (BEDAM), which runs using the Levy group’s molecular dynamics engine IMPACT. This methodology uses Hamiltonian replica exchange molecular dynamics simulations of the ligand-receptor complex in an implicit solvent model to construct a free energy path that connects the unbound and bound states of the ligand with the receptor.

Running BEDAM on IBM’s WCG
Differences between Phase I and Phase II

For FAH2, the concept of a batch has changed from 1,000-10,000 ligand-receptor complexes per batch to 1 complex per batch, and each workunit in that batch is running thousands of simulations of that complex which we call replicas. These replicas differ from each other in important ways, such as different energy function parameters and different starting conformations of the ligand and receptor.

Another key difference between Phase I and Phase II is that the simulation corresponding to one replica is broken up into several workunits that must be completed serially, i.e., the output of one workunit serves as input for the next workunit. Each batch generates tens of thousands of workunits that together form the simulations of hundreds or thousands of replica trajectories.
First computational experiments

Additionally, the distributed and heterogeneous nature of the volunteer grid imposes some constraints on how the simulations can be run. We’re in the process of learning how best to utilize this enormous resource by running each batch with two different simulation schema and several different analysis protocols. Doing so will allow us to refine our future simulation schema in order to maximize the impact of donated computing cycles. As volunteers ourselves, we do not want to see wasted computing cycles and are working with the constraints imposed by our dynamics engine, our forcefield model, and the nature of the WCG to find the optimal computing methodology for us and the volunteers.

The two simulation protocols currently being tested employ sampling techniques that differ from the way these simulations are run on clusters and supercomputers. The first is what we call “lambda scheduling” or “lambda cycling”, which is where throughout a replica’s simulation, key parameters associated with the coupling between ligand and receptor are cycled through a known set of values. This greatly helps to accelerate the sampling of the conformational space. Simulating multiple replicas with the same parameter changing schedule helps accelerate this sampling further.

The second simulation scheme is independent sampling in which no two replicas will ever have the same combination of energy function parameters, initial structures, etc., and these parameters are constant throughout all workunits pertaining to each replica’s trajectory. With both simulation schema, long simulation times are needed. We have developed analysis techniques which can combine the results from all of these different replica trajectories to produce estimates of the protein-ligand complex’s binding free energy.

HIV Integration (left) and HIV maturation/cleavage (right) as illustrated by David S. Goodsell © 2015, All Rights Reserved. Read more at the HIV Interaction and Viral Evolution Center.

See the full article here.

Please help promote STEM in your local schools.

Stem Education Coalition

World Community Grid (WCG) brings people together from across the globe to create the largest non-profit computing grid benefiting humanity. It does this by pooling surplus computer processing power. We believe that innovation combined with visionary scientific research and large-scale volunteerism can help make the planet smarter. Our success depends on like-minded individuals – like you.”

WCG projects run on BOINC software from UC Berkeley.

BOINC is a leader in the field(s) of Distributed Computing, Grid Computing and Citizen Cyberscience.BOINC is more properly the Berkeley Open Infrastructure for Network Computing.

BOINC WallPaper


“Download and install secure, free software that captures your computer’s spare power when it is on, but idle. You will then be a World Community Grid volunteer. It’s that simple!” You can download the software at either WCG or BOINC.

Please visit the project pages-

Rutgers Open Zika

WCG Help Stop TB
Help Stop TB
WCG Help Stop TB
Outsmart Ebola together

Outsmart Ebola Together

Mapping Cancer Markers

Uncovering Genome Mysteries
Uncovering Genome Mysteries

Say No to Schistosoma

GO Fight Against Malaria

Drug Search for Leishmaniasis

Computing for Clean Water

The Clean Energy Project

Discovering Dengue Drugs – Together

Help Cure Muscular Dystrophy

Help Fight Childhood Cancer

Help Conquer Cancer

Human Proteome Folding


World Community Grid is a social initiative of IBM Corporation
IBM Corporation

IBM – Smarter Planet