Rapid, accurate, precise and reproducible ligand-protein binding free energy prediction.

A central quantity of interest in molecular biology and medicine is the free energy of binding of a molecule to a target biomacromolecule. Until recently, the accurate prediction of binding affinity had been widely regarded as out of reach of theoretical methods owing to the lack of reproducibility of the available methods, not to mention their complexity, computational cost and time-consuming procedures. The lack of reproducibility stems primarily from the chaotic nature of classical molecular dynamics (MD) and the associated extreme sensitivity of trajectories to their initial conditions. Here, we review computational approaches for both relative and absolute binding free energy calculations, and illustrate their application to a diverse set of ligands bound to a range of proteins with immediate relevance in a number of medical domains. We focus on ensemble-based methods which are essential in order to compute statistically robust results, including two we have recently developed, namely thermodynamic integration with enhanced sampling and enhanced sampling of MD with an approximation of continuum solvent. Together, these form a set of rapid, accurate, precise and reproducible free energy methods. They can be used in real-world problems such as hit-to-lead and lead optimization stages in drug discovery, and in personalized medicine. These applications show that individual binding affinities equipped with uncertainty quantification may be computed in a few hours on a massive scale given access to suitable high-end computing resources and workflow automation. A high level of accuracy can be achieved using these approaches.

High-throughput binding affinity calculations at extreme scales.

BACKGROUND: Resistance to chemotherapy and molecularly targeted therapies is a major factor in limiting the effectiveness of cancer treatment. In many cases, resistance can be linked to genetic changes in target proteins, either pre-existing or evolutionarily selected during treatment. Key to overcoming this challenge is an understanding of the molecular determinants of drug binding. Using multi-stage pipelines of molecular simulations we can gain insights into the binding free energy and the residence time of a ligand, which can inform both stratified and personal treatment regimes and drug development. To support the scalable, adaptive and automated calculation of the binding free energy on high-performance computing resources, we introduce the High-throughput Binding Affinity Calculator (HTBAC). HTBAC uses a building block approach in order to attain both workflow flexibility and performance. RESULTS: We demonstrate close to perfect weak scaling to hundreds of concurrent multi-stage binding affinity calculation pipelines. This permits a rapid time-to-solution that is essentially invariant of the calculation protocol, size of candidate ligands and number of ensemble simulations. CONCLUSIONS: As such, HTBAC advances the state of the art of binding affinity calculations and protocols. HTBAC provides the platform to enable scientists to study a wide range of cancer drugs and candidate ligands in order to support personalized clinical decision making based on genome sequencing and drug discovery.

Rapid and Reliable Binding Affinity Prediction of Bromodomain Inhibitors: A Computational Study.

Binding free energies of bromodomain inhibitors are calculated with recently formulated approaches, namely ESMACS (enhanced sampling of molecular dynamics with approximation of continuum solvent) and TIES (thermodynamic integration with enhanced sampling). A set of compounds is provided by GlaxoSmithKline, which represents a range of chemical functionality and binding affinities. The predicted binding free energies exhibit a good Spearman correlation of 0.78 with the experimental data from the 3-trajectory ESMACS, and an excellent correlation of 0.92 from the TIES approach where applicable. Given access to suitable high end computing resources and a high degree of automation, we can compute individual binding affinities in a few hours with precisions no greater than 0.2 kcal/mol for TIES, and no larger than 0.34 and 1.71 kcal/mol for the 1- and 3-trajectory ESMACS approaches.

From base pair to bedside: molecular simulation and the translation of genomics to personalized medicine.

Despite the promises made that genomic sequencing would transform therapy by introducing a new era of personalized medicine, relatively few tangible breakthroughs have been made. This has led to the recognition that complex interactions at multiple spatial, temporal, and organizational levels may often combine to produce disease. Understanding this complexity requires that existing and future models are used and interpreted within a framework that incorporates knowledge derived from investigations at multiple levels of biological function. It also requires a computational infrastructure capable of dealing with the vast quantities of data generated by genomic approaches. In this review, we discuss the use of molecular modeling to generate quantitative and qualitative insights at the smallest scales of the systems biology hierarchy, how it can play an important role in the development of a systems understanding of disease and in the application of such knowledge to help discover new therapies and target existing ones on a personal level.

On the usability of grid middleware and security mechanisms.

Usability is an all too often neglected aspect of grid computing, although it is one of the principal factors militating against the widespread uptake of distributed computing. Many resource providers on a grid infrastructure deploy a standard middleware stack and expect users to invoke the default client tools for that middleware stack to access their resources. Unfortunately, many of these middleware client tools have been developed as an afterthought, and are widely considered difficult to use. Such tools typically require a user to interact with a machine, to stage data and launch jobs, and to use digital certificates. Our experience of working with grids over many years has led us to propose a new model of grid interaction, which we call the user-application interaction model. Similar considerations have also led us to develop environments that remove digital certificates from the user's experience, replacing them with familiar username and password authentication credentials. In this paper, we investigate the usability of this interaction model and its security system through a series of tests, which compare the usability of our systems with commonly deployed middleware tools using five usability metrics. Our middleware and security solutions are judged to be more usable than the systems in use by most of today's computational grids.

Audited credential delegation: a usable security solution for the virtual physiological human toolkit.

We present applications of audited credential delegation (ACD), a usable security solution for authentication, authorization and auditing in distributed virtual physiological human (VPH) project environments that removes the use of digital certificates from end-users' experience. Current security solutions are based on public key infrastructure (PKI). While PKI offers strong security for VPH projects, it suffers from serious usability shortcomings in terms of end-user acquisition and management of credentials which deter scientists from exploiting distributed VPH environments. By contrast, ACD supports the use of local credentials. Currently, a local ACD username-password combination can be used to access grid-based resources while Shibboleth support is underway. Moreover, ACD provides seamless and secure access to shared patient data, tools and infrastructure, thus supporting the provision of personalized medicine for patients, scientists and clinicians participating in e-health projects from a local to the widest international scale.

Petascale lattice-Boltzmann studies of amphiphilic cubic liquid crystalline materials in a globally distributed high-performance computing and visualization environment.

We present very large-scale rheological studies of self-assembled cubic gyroid liquid crystalline phases in ternary mixtures of oil, water and amphiphilic species performed on petascale supercomputers using the lattice-Boltzmann method. These nanomaterials have found diverse applications in materials science and biotechnology, for example, in photovoltaic devices and protein crystallization. They are increasingly gaining importance as delivery vehicles for active agents in pharmaceuticals, personal care products and food technology. In many of these applications, the self-assembled structures are subject to flows of varying strengths and we endeavour to understand their rheological response with the objective of eventually predicting it under given flow conditions. Computationally, our lattice-Boltzmann simulations of ternary fluids are inherently memory- and data-intensive. Furthermore, our interest in dynamical processes necessitates remote visualization and analysis as well as the associated transfer and storage of terabytes of time-dependent data. These simulations are distributed on a high-performance grid infrastructure using the application hosting environment; we employ a novel parallel in situ visualization approach which is particularly suited for such computations on petascale resources. We present computational and I/O performance benchmarks of our application on three different petascale systems.

Automated molecular simulation based binding affinity calculator for ligand-bound HIV-1 proteases.

The successful application of high throughput molecular simulations to determine biochemical properties would be of great importance to the biomedical community if such simulations could be turned around in a clinically relevant timescale. An important example is the determination of antiretroviral inhibitor efficacy against varying strains of HIV through calculation of drug-protein binding affinities. We describe the Binding Affinity Calculator (BAC), a tool for the automated calculation of HIV-1 protease-ligand binding affinities. The tool employs fully atomistic molecular simulations alongside the well established molecular mechanics Poisson-Boltzmann solvent accessible surface area (MMPBSA) free energy methodology to enable the calculation of the binding free energy of several ligand-protease complexes, including all nine FDA approved inhibitors of HIV-1 protease and seven of the natural substrates cleaved by the protease. This enables the efficacy of these inhibitors to be ranked across several mutant strains of the protease relative to the wildtype. BAC is a tool that utilizes the power provided by a computational grid to automate all of the stages required to compute free energies of binding: model preparation, equilibration, simulation, postprocessing, and data-marshaling around the generally widely distributed compute resources utilized. Such automation enables the molecular dynamics methodology to be used in a high throughput manner not achievable by manual methods. This paper describes the architecture and workflow management of BAC and the function of each of its components. Given adequate compute resources, BAC can yield quantitative information regarding drug resistance at the molecular level within 96 h. Such a timescale is of direct clinical relevance and can assist in decision support for the assessment of patient-specific optimal drug treatment and the subsequent response to therapy for any given genotype.

Patient-specific simulation as a basis for clinical decision-making.

Patient-specific medical simulation holds the promise of determining tailored medical treatment based on the characteristics of an individual patient (for example, using a genotypic assay of a sequence of DNA). Decision-support systems based on patient-specific simulation can potentially revolutionize the way that clinicians plan courses of treatment for various conditions, ranging from viral infections to arterial abnormalities. Basing medical decisions on the results of simulations that use models derived from data specific to the patient in question means that the effectiveness of a range of potential treatments can be assessed before they are actually administered, preventing the patient from experiencing unnecessary or ineffective treatments. We illustrate the potential for patient-specific simulation by first discussing the scale of the evolving international grid infrastructure that is now available to underpin such applications. We then consider two case studies, one concerned with the treatment of patients with HIV/AIDS and the other addressing neuropathologies associated with the intracranial vasculature. Such patient-specific medical simulations require access to both appropriate patient data and the computational resources on which to perform potentially very large simulations. Computational infrastructure providers need to furnish access to a wide range of different types of resource, typically made available through heterogeneous computational grids, and to institute policies that facilitate the performance of patient-specific simulations on those resources. To support these kinds of simulations, where life and death decisions are being made, computational resource providers must give urgent priority to such jobs, for example by allowing them to pre-empt the queue on a machine and run straight away. We describe systems that enable such priority computing.