The truncated conjugate gradient (TCG), a non-iterative/fixed-cost strategy for computing polarization in molecular dynamics: Fast evaluation of analytical forces

In a recent paper [F. Aviat et al., J. Chem. Theory Comput. 13, 180-190 (2017)], we proposed the Truncated Conjugate Gradient (TCG) approach to compute the polarization energy and forces in polarizable molecular simulations. The method consists in truncating the conjugate gradient algorithm at a fixed predetermined order leading to a fixed computational cost and can thus be considered "non-iterative." This gives the possibility to derive analytical forces avoiding the usual energy conservation (i.e., drifts) issues occurring with iterative approaches. A key point concerns the evaluation of the analytical gradients, which is more complex than that with a usual solver. In this paper, after reviewing the present state of the art of polarization solvers, we detail a viable strategy for the efficient implementation of the TCG calculation. The complete cost of the approach is then measured as it is tested using a multi-time step scheme and compared to timings using usual iterative approaches. We show that the TCG methods are more efficient than traditional techniques, making it a method of choice for future long molecular dynamics simulations using polarizable force fields where energy conservation matters. We detail the various steps required for the implementation of the complete method by software developers.

Preface: Special Topic: From Quantum Mechanics to Force Fields

This Special Topic issue entitled "From Quantum Mechanics to Force Fields" is dedicated to the ongoing efforts of the theoretical chemistry community to develop a new generation of accurate force fields based on data from high-level electronic structure calculations and to develop faster electronic structure methods for testing and designing force fields as well as for carrying out simulations. This issue includes a collection of 35 original research articles that illustrate recent theoretical advances in the field. It provides a timely snapshot of recent developments in the generation of approaches to enable more accurate molecular simulations of processes important in chemistry, physics, biophysics, and materials science.

QM/MM Simulations with the Gaussian Electrostatic Model: A Density-based Polarizable Potential

The use of advanced polarizable potentials in quantum mechanical/molecular mechanical (QM/MM) simulations has been shown to improve the overall accuracy of the calculation. We have developed a density-based potential called the Gaussian electrostatic model (GEM), which has been shown to provide very accurate environments for QM wave functions in QM/MM. In this contribution we present a new implementation of QM/GEM that extends our implementation to include all components (Coulomb, exchange-repulsion, polarization, and dispersion) for the total intermolecular interaction energy in QM/MM calculations, except for the charge-transfer term. The accuracy of the method is tested using a subset of water dimers from the water dimer potential energy surface reported by Babin et al. ( J. Chem. Theory Comput. 2013 9, 5395-5403). Additionally, results of the new implementation are contrasted with results obtained with the classical AMOEBA potential. Our results indicate that GEM provides an accurate MM environment with average root-mean-square error <0.15 kcal/mol for every intermolecular interaction energy component compared with SAPT2+3/aug-cc-pVTZ reference calculations.

AMOEBA Polarizable Force Field Parameters of the Heme Cofactor in Its Ferrous and Ferric Forms

We report the first parameters of the heme redox cofactors for the polarizable AMOEBA force field in both the ferric and ferrous forms. We consider two types of complexes, one with two histidine side chains as axial ligands and one with a histidine and a methionine side chain as ligands. We have derived permanent multipoles from second-order Møller-Plesset perturbation theory (MP2). The sets of parameters have been validated in a first step by comparison of AMOEBA interaction energies of heme and a collection of biologically relevant molecules with MP2 and Density Functional Theory (DFT) calculations. In a second validation step, we consider interaction energies with large aggregates comprising around 80 H₂O molecules. These calculations are repeated for 30 structures extracted from semiempirical PM7 DM simulations. Very encouraging agreement is found between DFT and the AMOEBA force field, which results from an accurate treatment of electrostatic interactions. We finally report long (10 ns) MD simulations of cytochromes in two redox states with AMOEBA testing both the 2003 and 2014 AMOEBA water models. These simulations have been carried out with the TINKER-HP (High Performance) program. In conclusion, owing to their ubiquity in biology, we think the present work opens a wide array of applications of the polarizable AMOEBA force field on hemeproteins.

AMOEBA Polarizable Atomic Multipole Force Field for Nucleic Acids

The AMOEBA polarizable atomic multipole force field for nucleic acids is presented. Valence and electrostatic parameters were determined from high-level quantum mechanical data, including structures, conformational energy, and electrostatic potentials, of nucleotide model compounds. Previously derived parameters for the phosphate group and nucleobases were incorporated. A total of over 35 μs of condensed-phase molecular dynamics simulations of DNA and RNA molecules in aqueous solution and crystal lattice were performed to validate and refine the force field. The solution and/or crystal structures of DNA B-form duplexes, RNA duplexes, and hairpins were captured with an average root-mean-squared deviation from NMR structures below or around 2.0 Å. Structural details, such as base pairing and stacking, sugar puckering, backbone and χ-torsion angles, groove geometries, and crystal packing interfaces, agreed well with NMR and/or X-ray. The interconversion between A- and B-form DNAs was observed in ethanol-water mixtures at 328 K. Crystal lattices of B- and Z-form DNA and A-form RNA were examined with simulations. For the RNA tetraloop, single strand tetramers, and HIV TAR with 29 residues, the simulated conformational states, 3 J-coupling, nuclear Overhauser effect, and residual dipolar coupling data were compared with NMR results. Starting from a totally unstacked/unfolding state, the rCAAU tetranucleotide was folded into A-form-like structures during ∼1 μs molecular dynamics simulations.

The Independent Gradient Model: A New Approach for Probing Strong and Weak Interactions in Molecules from Wave Function Calculations

Extraction of the chemical interaction signature from local descriptors based on electron density (ED) is still a fruitful field of development in chemical interpretation. In a previous work that used promolecular ED (frozen ED), the new descriptor, δg , was defined. It represents the difference between a virtual upper limit of the ED gradient (∇ρIGM , IGM=independent gradient model) that represents a noninteracting system and the true ED gradient (∇ρ ). It can be seen as a measure of electron sharing brought by ED contragradience. A compelling feature of this model is to provide an automatic workflow that extracts the signature of interactions between selected groups of atoms. As with the noncovalent interaction (NCI) approach, it provides chemists with a visual understanding of the interactions present in chemical systems. ∇ρIGM is achieved simply by using absolute values upon summing the individual gradient contributions that make up the total ED gradient. Hereby, we extend this model to relaxed ED calculated from a wave function. To this end, we formulated gradient-based partitioning (GBP) to assess the contribution of each orbital to the total ED gradient. We highlight these new possibilities across two prototypical examples of organic chemistry: the unconventional hexamethylbenzene dication, with a hexa-coordinated carbon atom, and β-thioaminoacrolein. It will be shown how a bond-by-bond picture can be obtained from a wave function, which opens the way to monitor specific interactions along reaction paths.

An optimized charge penetration model for use with the AMOEBA force field

The principal challenge of using classical physics to model biomolecular interactions is capturing the nature of short-range interactions that drive biological processes from nucleic acid base stacking to protein-ligand binding. In particular most classical force fields suffer from an error in their electrostatic models that arises from an ability to account for the overlap between charge distributions occurring when molecules get close to each other, known as charge penetration. In this work we present a simple, physically motivated model for including charge penetration in the AMOEBA (Atomic Multipole Optimized Energetics for Biomolecular Applications) force field. With a function derived from the charge distribution of a hydrogen-like atom and a limited number of parameters, our charge penetration model dramatically improves the description of electrostatics at short range. On a database of 101 biomolecular dimers, the charge penetration model brings the error in the electrostatic interaction energy relative to the ab initio SAPT electrostatic interaction energy from 13.4 kcal mol^-1 to 1.3 kcal mol^-1. The model is shown not only to be robust and transferable for the AMOEBA model, but also physically meaningful as it universally improves the description of the electrostatic potential around a given molecule.

Tinker-HP: a massively parallel molecular dynamics package for multiscale simulations of large complex systems with advanced point dipole polarizable force fields

We present Tinker-HP, a massively MPI parallel package dedicated to classical molecular dynamics (MD) and to multiscale simulations, using advanced polarizable force fields (PFF) encompassing distributed multipoles electrostatics. Tinker-HP is an evolution of the popular Tinker package code that conserves its simplicity of use and its reference double precision implementation for CPUs. Grounded on interdisciplinary efforts with applied mathematics, Tinker-HP allows for long polarizable MD simulations on large systems up to millions of atoms. We detail in the paper the newly developed extension of massively parallel 3D spatial decomposition to point dipole polarizable models as well as their coupling to efficient Krylov iterative and non-iterative polarization solvers. The design of the code allows the use of various computer systems ranging from laboratory workstations to modern petascale supercomputers with thousands of cores. Tinker-HP proposes therefore the first high-performance scalable CPU computing environment for the development of next generation point dipole PFFs and for production simulations. Strategies linking Tinker-HP to Quantum Mechanics (QM) in the framework of multiscale polarizable self-consistent QM/MD simulations are also provided. The possibilities, performances and scalability of the software are demonstrated via benchmarks calculations using the polarizable AMOEBA force field on systems ranging from large water boxes of increasing size and ionic liquids to (very) large biosystems encompassing several proteins as well as the complete satellite tobacco mosaic virus and ribosome structures. For small systems, Tinker-HP appears to be competitive with the Tinker-OpenMM GPU implementation of Tinker. As the system size grows, Tinker-HP remains operational thanks to its access to distributed memory and takes advantage of its new algorithmic enabling for stable long timescale polarizable simulations. Overall, a several thousand-fold acceleration over a single-core computation is observed for the largest systems. The extension of the present CPU implementation of Tinker-HP to other computational platforms is discussed.

A Simple Isomerization of the Purine Scaffold of a Kinase Inhibitor, Roscovitine, Affords a Four- to Seven-Fold Enhancement of Its Affinity for Four CDKs. Could This Be Traced Back to Conjugation-Induced Stiffenings/Loosenings of Rotational Barriers?

Roscovitine is an antitumor purine inhibitor of cyclin-dependent kinase CDK5, for which it displays submicromolar affinity. It reached phase IIb clinical trials in 2007. The search for analogues with improved kinase affinities led recently to an isomer, finisterine, having a nearly 10-fold greater affinity for both CDK5 and CDK9. It solely differs by the displacement of one nitrogen atom in the purine ring, from position 6 to position 9. This has no incidence on the intermolecular interaction of either drug with the neighboring sites that anchor the ring in the recognition site. Quantum chemistry calculations combined with conformational and topological analyses of the impact of the purine ring isomerization of roscovitine and finisterine on its conformational stability show that the modified electronic conjugation, on the other hand, results in a stiffening of the rotational barrier around the extracyclic C-NH bond of finisterine, vicinal to N9, and to which an aryl ring is connected, along with a loosening of the barrier around an extracyclic C6-C bond connecting to a shorter, hydrophobic arm. The first effect is proposed to lead to a lesser hydration entropy of solvation in the case of finisterine, thus to a facilitated desolvation term in the overall energy balances.

The inhibition process of HIV-1 integrase by diketoacids molecules: Understanding the factors governing the better efficiency of dolutegravir

The Human Immunodeficiency Virus-1 integrase is responsible for the covalent insertion of a newly synthesized double-stranded viral DNA into the host cells, and is an emerging target for antivirus drug design. Raltegravir (RAL) and elvitegravir (EVG) are the first two integrase strand transfer inhibitors used in therapy. However, treated patients eventually develop detrimental resistance mutations. By contrast, a recently approved drug, dolutegravir (DTG), presents a high barrier to resistance. This study aims to understand the increased efficiency of DTG upon focusing on its interaction properties with viral DNA. The results showed DTG to be involved in more extended interactions with viral DNA than EVG. Such interactions involve the halobenzene and scaffold of DTG and EVG and bases 5'G^-43', 3'A³5'and 3'C⁴5'.

Tinker-OpenMM: Absolute and relative alchemical free energies using AMOEBA on GPUs

The capabilities of the polarizable force fields for alchemical free energy calculations have been limited by the high computational cost and complexity of the underlying potential energy functions. In this work, we present a GPU-based general alchemical free energy simulation platform for polarizable potential AMOEBA. Tinker-OpenMM, the OpenMM implementation of the AMOEBA simulation engine has been modified to enable both absolute and relative alchemical simulations on GPUs, which leads to a ∼200-fold improvement in simulation speed over a single CPU core. We show that free energy values calculated using this platform agree with the results of Tinker simulations for the hydration of organic compounds and binding of host-guest systems within the statistical errors. In addition to absolute binding, we designed a relative alchemical approach for computing relative binding affinities of ligands to the same host, where a special path was applied to avoid numerical instability due to polarization between the different ligands that bind to the same site. This scheme is general and does not require ligands to have similar scaffolds. We show that relative hydration and binding free energy calculated using this approach match those computed from the absolute free energy approach. © 2017 Wiley Periodicals, Inc.

Channeling through Two Stacked Guanine Quartets of One and Two Alkali Cations in the Li+, Na+, K+, and Rb+ Series. Assessment of the Accuracy of the SIBFA Anisotropic Polarizable Molecular Mechanics Potential

Stacking of guanine quartets (GQs) can trigger the formation of DNA or RNA quadruple helices, which play numerous biochemical roles. The GQs are stabilized by alkali cations, mainly K⁺ and Na⁺, which can reside in, or channel through, the central axis of the GQ stems. Further, ion conduction through GQ wires can be leveraged for nanochemistry applications. G-quadruplex systems have been extensively studied by classical molecular dynamics (MD) simulations using pair-additive force fields or by quantum-chemical (QC) calculations. However, the non-polarizable force fields are very approximate, while QC calculations lack the necessary sampling. Thus, ultimate description of GQ systems would require long-enough simulations using advanced polarizable molecular mechanics (MM). However, to perform such calculations, it is first mandatory to evaluate the method's accuracy using benchmark QC. We report such an evaluation for SIBFA polarizable MM, bearing on the channeling (movement) of an alkali cation (Li⁺, Na⁺, K⁺, or Rb⁺) along the axis of two stacked G quartets interacting with either one or two ions. The QC energy profiles display markedly different features depending upon the cation but can be retrieved in the majority of cases by the SIBFA profiles. An appropriate balance of first-order (electrostatic and short-range repulsion) and second-order (polarization, charge-transfer, and dispersion) contributions within ΔE is mandatory. With two cations in the channel, the relative weights of the second-order contributions increase steadily upon increasing the ion size. In the G8 complexes with two K⁺ or two Rb⁺ cations, the sum of polarization and charge-transfer exceeds the first order terms for all ion positions.

Truncated Conjugate Gradient: An Optimal Strategy for the Analytical Evaluation of the Many-Body Polarization Energy and Forces in Molecular Simulations

We introduce a new class of methods, denoted as Truncated Conjugate Gradient(TCG), to solve the many-body polarization energy and its associated forces in molecular simulations (i.e. molecular dynamics (MD) and Monte Carlo). The method consists in a fixed number of Conjugate Gradient (CG) iterations. TCG approaches provide a scalable solution to the polarization problem at a user-chosen cost and a corresponding optimal accuracy. The optimality of the CG-method guarantees that the number of the required matrix-vector products are reduced to a minimum compared to other iterative methods. This family of methods is non-empirical, fully adaptive, and provides analytical gradients, avoiding therefore any energy drift in MD as compared to popular iterative solvers. Besides speed, one great advantage of this class of approximate methods is that their accuracy is systematically improvable. Indeed, as the CG-method is a Krylov subspace method, the associated error is monotonically reduced at each iteration. On top of that, two improvements can be proposed at virtually no cost: (i) the use of preconditioners can be employed, which leads to the Truncated Preconditioned Conjugate Gradient (TPCG); (ii) since the residual of the final step of the CG-method is available, one additional Picard fixed point iteration (“peek”), equivalent to one step of Jacobi Over Relaxation (JOR) with relaxation parameter ω, can be made at almost no cost. This method is denoted by TCG-n(ω). Black-box adaptive methods to find good choices of ω are provided and discussed. Results show that TPCG-3(ω) is converged to high accuracy (a few kcal/mol) for various types of systems including proteins and highly charged systems at the fixed cost of four matrix-vector products: three CG iterations plus the initial CG descent direction. Alternatively, T(P)CG-2(ω) provides robust results at a reduced cost (three matrix-vector products) and offers new perspectives for long polarizable MD as a production algorithm. The T(P)CG-1(ω) level provides less accurate solutions for inhomogeneous systems, but its applicability to well-conditioned problems such as water is remarkable, with only two matrix-vector product evaluations.

Polarizable molecular dynamics in a polarizable continuum solvent

We present, for the first time, scalable polarizable molecular dynamics (MD) simulations within a polarizable continuum solvent with molecular shape cavities and exact solution of the mutual polarization. The key ingredients are a very efficient algorithm for solving the equations associated with the polarizable continuum, in particular, the domain decomposition Conductor-like Screening Model (ddCOSMO), which involves a rigorous coupling of the continuum with the polarizable force field achieved through a robust variational formulation and an effective strategy to solve the coupled equations. The coupling of ddCOSMO with nonvariational force fields, including AMOEBA, is also addressed. The MD simulations are feasible, for real-life systems, on standard cluster nodes; a scalable parallel implementation allows for further acceleration in the context of a newly developed module in Tinker, named Tinker-HP. NVE simulations are stable, and long-term energy conservation can be achieved. This paper is focused on the methodological developments, the analysis of the algorithm, and the stability of the simulations; a proof-of-concept application is also presented to attest to the possibilities of this newly developed technique.

LICHEM: A QM/MM program for simulations with multipolar and polarizable force fields

We introduce an initial implementation of the LICHEM software package. LICHEM can interface with Gaussian, PSI4, NWChem, TINKER, and TINKER-HP to enable QM/MM calculations using multipolar/polarizable force fields. LICHEM extracts forces and energies from unmodified QM and MM software packages to perform geometry optimizations, single-point energy calculations, or Monte Carlo simulations. When the QM and MM regions are connected by covalent bonds, the pseudo-bond approach is employed to smoothly transition between the QM region and the polarizable force field. A series of water clusters and small peptides have been employed to test our initial implementation. The results obtained from these test systems show the capabilities of the new software and highlight the importance of including explicit polarization. © 2016 Wiley Periodicals, Inc.

Scalable improvement of SPME multipolar electrostatics in anisotropic polarizable molecular mechanics using a general short-range penetration correction up to quadrupoles

We propose a general coupling of the Smooth Particle Mesh Ewald SPME approach for distributed multipoles to a short-range charge penetration correction modifying the charge-charge, charge-dipole and charge-quadrupole energies. Such an approach significantly improves electrostatics when compared to ab initio values and has been calibrated on Symmetry-Adapted Perturbation Theory reference data. Various neutral molecular dimers have been tested and results on the complexes of mono- and divalent cations with a water ligand are also provided. Transferability of the correction is adressed in the context of the implementation of the AMOEBA and SIBFA polarizable force fields in the TINKER-HP software. As the choices of the multipolar distribution are discussed, conclusions are drawn for the future penetration-corrected polarizable force fields highlighting the mandatory need of non-spurious procedures for the obtention of well balanced and physically meaningful distributed moments. Finally, scalability and parallelism of the short-range corrected SPME approach are addressed, demonstrating that the damping function is computationally affordable and accurate for molecular dynamics simulations of complex bio- or bioinorganic systems in periodic boundary conditions.