Transferable Neural Network Potential Energy Surfaces for Closed-Shell Organic Molecules: Extension to Ions

Transferable high dimensional neural network potentials (HDNNP) have shown great promise as an avenue to increase the accuracy and domain of applicability of existing atomistic force fields for organic systems relevant to life science. We have previously reported such a potential (Schrödinger-ANI) that has broad coverage of druglike molecules. We extend that work here to cover ionic and zwitterionic druglike molecules expected to be relevant to drug discovery research activities. We report a novel HDNNP architechture, which we call QRNN, that predicts atomic charges and uses these charges as descriptors in an energy model which delivers conformational energies within chemical accuracy when measured against the reference theory it is trained to. Further, we find that delta learning based on a semi-empirical level of theory approximately halves the errors. We test the models on torsion energy profiles, relative conformational energies, geometric parameters and relative tautomer errors.

An asymmetric magnetic-coupled bending-torsion piezoelectric energy harvester: modeling and experimental investigation

This paper presents a detailed investigation on an asymmetric magnetic-coupled bending-torsion piezoelectric energy harvester based on harmonic excitation. There is an eccentricity between the shape center of moving magnets and the axis of the piezoelectric beam, which results in the bending and torsion simultaneously in working condition. The distributed mathematical model is derived from the energy method to describe the dynamic characteristics of the harvester, and the correctness of the model is verified by experiments. To further demonstrate the improvement performance of the proposed energy harvester, the bending-torsion energy harvester (i.e. magnetic-coupled was not configured) is experimented and compared. The theoretical and experimental results indicate that the average power increases about 300% but the resonance frequency decreases approximately 2 Hz comparing to the harvester without magnetic-coupled. According to the characteristic of distributed parameter model, the magnetic force and the size of the piezoelectric beam are investigated respectively. And the lumped-parameter model is introduced to analyze the steady-state characteristic. Accordingly, this paper provides a feasible method to improve performance for piezoelectric energy harvester.

Numerical Simulations of Detections, Experiments and Magnetic Field Hall Effect Analysis to Field Torsion

Field torsion models are considered from the experiments realized in electronic-dynamical devices and magnetic censoring of a Hall Effect sensor to detect torsion under electrical restricted conditions and space geometry. In this last point, are obtained 2D and 3D-models of torsion energy, which enclose the field theory concepts related with torsion, and open several possibilities of re-interpretations that can be useful in technological applications in the future. Likewise, are considered some measurements that evidence the torsion as field observable. Through geometrical models obtained from theorems and other results are demonstrated the conjectures required to understanding of torsion, as a geometrical and physics invariant most important in the deep study of the Universe. Also, applications in astrophysics and cosmology of these geometrical models are obtained to show Universe phenomena understudy of torsion.

Optimization of chromatographic separation of aripiprazole and impurities: Quantitative structure-retention relationship approach

New optimization strategy based on mixed Quantitative Structure-Retention Relationship (QSRR) model was proposed for improving the RP-HPLC separation of aripiprazole and its impurities (IMP A-E). Firstly, experimental parameters (EPs) (mobile phase composition and flow rate) were varied according to Box-Behnken Design and afterwards, artificial neural network (ANN) as QSRR model was built correlating EPs and selected molecular descriptors (ovality, torsion energy and non-1,4-Van der Waals energy) with analytes log-transformed retention time. Values of root mean square error (RMSE) were used for ANNs quality estimation (0.0227, 0.0191 and 0.0230 for training, verification and test set, respectively). Separations of critical peak pairs on chromatogram (IMP A-B and IMP D-C) were optimized using ANNs for which EPs served as inputs and log-transformed separation criteria s as outputs. They were validated applying leave-one-out cross-validation (RMSE values 0.065 and 0.056, respectively). Obtained ANNs were used for plotting response surfaces upon which analyses chromatographic conditions resulting in optimal analytes retention behaviour and optimal values of separation criteria s were defined and they comprised of 54 % of methanol at the beginning and 79 % of methanol at the end of gradient elution programme with mobile phase flow rate of 460 ?L min-1.

TorsionNet: A Deep Neural Network to Rapidly Predict Small Molecule Torsion Energy Profiles with the Accuracy of Quantum Mechanics

<p></p><p>TorsionNet: A Deep Neural Network to Rapidly Predict Small Molecule Torsion Energy Profiles with the Accuracy of Quantum Mechanics </p> <p> </p> <p>Brajesh K. Rai^*,1, Vishnu Sresht¹, Qingyi Yang², Ray Unwalla², Meihua Tu², Alan M. Mathiowetz², and Gregory A. Bakken³</p> <p>¹Simulation and Modeling Sciences and ²Medicine Design, Pfizer Worldwide Research Development and Medical, 610 Main Street, Cambridge, Massachusetts 02139, United States</p> <p>³Digital, Pfizer, Eastern Point Road, Groton, Connecticut 06340, United States</p> <p> </p> <p> </p> <p><b>ABSTRACT</b><b> </b><b></b></p> <p>Fast and accurate assessment of small molecule dihedral energetics is crucial for molecular design and optimization in medicinal chemistry. Yet, accurate prediction of torsion energy profiles remains a challenging task as current molecular mechanics methods are limited by insufficient coverage of druglike chemical space and accurate quantum mechanical (QM) methods are too expensive. To address this limitation, we introduce TorsionNet, a deep neural network (DNN) model specifically developed to predict small molecule torsion energy profiles with QM-level accuracy. We applied active learning to identify nearly 50k fragments (with elements H, C, N, O, F, S, and Cl) that maximized the coverage of our corporate library and leveraged massively parallel cloud computing resources to perform DFT torsion scan of these fragments, generating a training dataset of 1.2 million DFT energies. By training TorsionNet on this dataset, we obtain a model that can rapidly predict the torsion energy profile of typical druglike fragments with DFT-level accuracy. Importantly, our method also provides a direct estimate of the uncertainty in the predicted profiles without any additional calculations. In this report, we show that TorsionNet can reliably identify the preferred dihedral geometries observed in crystal structures. We also present practical applications of TorsionNet that demonstrate how consideration of DNN-based strain energy leads to substantial improvement in existing lead discovery and design workflows. A benchmark dataset (TorsionNet500) comprising 500 chemically diverse fragments with DFT torsion profiles (12k DFT-optimized geometries and energies) has been created and is made freely available.</p><br><p></p>

TorsionNet: A Deep Neural Network to Rapidly Predict Small Molecule Torsion Energy Profiles with the Accuracy of Quantum Mechanics

<p></p><p>TorsionNet: A Deep Neural Network to Rapidly Predict Small Molecule Torsion Energy Profiles with the Accuracy of Quantum Mechanics </p> <p> </p> <p>Brajesh K. Rai^*,1, Vishnu Sresht¹, Qingyi Yang², Ray Unwalla², Meihua Tu², Alan M. Mathiowetz², and Gregory A. Bakken³</p> <p>¹Simulation and Modeling Sciences and ²Medicine Design, Pfizer Worldwide Research Development and Medical, 610 Main Street, Cambridge, Massachusetts 02139, United States</p> <p>³Digital, Pfizer, Eastern Point Road, Groton, Connecticut 06340, United States</p> <p> </p> <p> </p> <p><b>ABSTRACT</b><b> </b><b></b></p> <p>Fast and accurate assessment of small molecule dihedral energetics is crucial for molecular design and optimization in medicinal chemistry. Yet, accurate prediction of torsion energy profiles remains a challenging task as current molecular mechanics methods are limited by insufficient coverage of druglike chemical space and accurate quantum mechanical (QM) methods are too expensive. To address this limitation, we introduce TorsionNet, a deep neural network (DNN) model specifically developed to predict small molecule torsion energy profiles with QM-level accuracy. We applied active learning to identify nearly 50k fragments (with elements H, C, N, O, F, S, and Cl) that maximized the coverage of our corporate library and leveraged massively parallel cloud computing resources to perform DFT torsion scan of these fragments, generating a training dataset of 1.2 million DFT energies. By training TorsionNet on this dataset, we obtain a model that can rapidly predict the torsion energy profile of typical druglike fragments with DFT-level accuracy. Importantly, our method also provides a direct estimate of the uncertainty in the predicted profiles without any additional calculations. In this report, we show that TorsionNet can reliably identify the preferred dihedral geometries observed in crystal structures. We also present practical applications of TorsionNet that demonstrate how consideration of DNN-based strain energy leads to substantial improvement in existing lead discovery and design workflows. A benchmark dataset (TorsionNet500) comprising 500 chemically diverse fragments with DFT torsion profiles (12k DFT-optimized geometries and energies) has been created and is made freely available.</p><br><p></p>

Capturing non-local through-bond effects when fragmenting molecules for quantum chemical torsion scans

Accurate molecular mechanics force fields for small molecules are essential for predicting protein-ligand binding affinities in drug discovery and understanding the biophysics of biomolecular systems. Torsion potentials derived from quantum chemical (QC) calculations are critical for determining the conformational distributions of small molecules, but are computationally expensive and scale poorly with molecular size. To reduce computational cost and avoid the complications of distal through-space intramolecular interactions, molecules are generally fragmented into smaller entities to carry out QC torsion scans. However, torsion potentials, particularly for conjugated bonds, can be strongly affected by through-bond chemistry distal to the torsion itself. Poor fragmentation schemes have the potential to significantly disrupt electronic properties in the region around the torsion by removing important, distal chemistries, leading to poor representation of the parent molecule’s chemical environment and the resulting torsion energy profile. Here we show that a rapidly computable quantity, the fractional Wiberg bond order (WBO), is a sensitive reporter on whether the chemical environment around a torsion has been disrupted. We show that the WBO can be used as a surrogate to assess the robustness of fragmentation schemes and identify conjugated bond sets. We use this concept to construct a validation set by exhaustively fragmenting a set of druglike organic molecules and examine their corresponding WBO distributions derived from accessible conformations that can be used to evaluate fragmentation schemes. To illustrate the utility of the WBO in assessing fragmentation schemes that preserve the chemical environment, we propose a new fragmentation scheme that uses rapidly-computable AM1 WBOs, which are available essentially for free as part of standard AM1-BCC partial charge assignment. This approach can simultaneously maximize the chemical equivalency of the fragment and the substructure in the larger molecule while minimizing fragment size to accelerate QC torsion potential computation for small molecules and reducing undesired through-space steric interactions.

Analytical modelling of spatial deformation pathways in planar and spatial shallow bistable arches

We analyse spatial bistable arches and present an analytical model incorporating axial, two transverse bending and torsion energy components. We extend the St. Venant and Michell relationship used in flexural-torsional buckling of planar arches and use it in modelling spatial arches. We study deformation pathways in spatial arches and their effect on critical characteristics of bistability such as back and forth switching forces, and the distance travelled by a point of the arch. We show that not considering spatial deformation leads to incorrect inferences concerning the bistability of planar arches too. Thus, this model serves as a generalized framework for the existing analysis on planar arches since they belong to a subset of spatial arches. Additionally, the effects of eccentric loading on spatial deformations are explored for arches with a range of as-fabricated shapes and boundary conditions, and the results are validated with finite-element analysis.

A new lipid force field (FUJI)

To explore inhomogeneous and anisotropic systems such as lipid bilayers, the Lennard-Jones particle mesh Ewald (LJ-PME) method was applied without a traditional isotropic dispersion correction. As the popular AMBER and CHARMM lipid force fields were developed using a cutoff scheme, their lipid bilayers unacceptably shrank when using LJ-PME method. A new lipid force field (FUJI) was developed on the basis of the AMBER force field scheme including the Lipid14 van der Waals parameters. Point charges were calculated by the restrained electrostatic potentials of many lipid conformers. The torsion energy profiles were calculated by high level ab initio molecular orbitals (LCCSD(T)/Aug-cc-pVTZ//LMP2/Aug-cc-pVTZ); then, the molecular mechanical dihedral parameters were derived by means of a fast Fourier transform. Incorporating these parameters into a new lipid force field without any fittings to experimental data, desirable lipid characteristics such as the area per lipid and lateral diffusion coefficients were obtained by GROMACS molecular dynamics simulations using the LJ-PME method and hydrogen virtual sites. The stability and structures of large membranes with undulatory fluctuations were studied by a multidrug efflux transporter (AcrABZ-TolC) with inner and outer membranes.

Dynamic calorimetry and XRD studies of the nematic and twist-bend nematic phase transitions in a series of dimers with increasing spacer length

A modulated DSC study of bent dimesogens with (CH2)n spacers n = 5–11 showed that the enthalpy of the ordinary nematic is lowest for n = 11 due to the lowest C–C torsion energy needed to straighten the dimer, causing near disappearance of twist-bend-nematic to nematic transition enthalpy.