Training atomic neural networks using fragment-based data generated in virtual reality.

The ability to understand and engineer molecular structures relies on having accurate descriptions of the energy as a function of atomic coordinates. Here, we outline a new paradigm for deriving energy functions of hyperdimensional molecular systems, which involves generating data for low-dimensional systems in virtual reality (VR) to then efficiently train atomic neural networks (ANNs). This generates high-quality data for specific areas of interest within the hyperdimensional space that characterizes a molecule's potential energy surface (PES). We demonstrate the utility of this approach by gathering data within VR to train ANNs on chemical reactions involving fewer than eight heavy atoms. This strategy enables us to predict the energies of much higher-dimensional systems, e.g., containing nearly 100 atoms. Training on datasets containing only 15k geometries, this approach generates mean absolute errors around 2 kcal mol-1. This represents one of the first times that an ANN-PES for a large reactive radical has been generated using such a small dataset. Our results suggest that VR enables the intelligent curation of high-quality data, which accelerates the learning process.

Guidelines for Recurrent Neural Network Transfer Learning-Based Molecular Generation of Focused Libraries.

Deep learning approaches have become popular in recent years in the field of de novo molecular design. While a variety of different methods are available, it is still a challenge to assess and compare their performance. A particularly promising approach for automated drug design is to use recurrent neural networks (RNNs) as SMILES generators and train them with the learning procedure called "transfer learning". This involves first training the initial model on a large generic data set of molecules to learn the general syntax of SMILES, followed by fine-tuning on a smaller set of molecules, coming from, e.g., a lead optimization program. To create a well-performing transfer learning application which can be automated, it is important to understand how the size of the second data set affects the training process. In addition, extensive postfiltering using similarity metrics of the molecules generated after transfer learning should be avoided, as it can introduce new biases toward the selection of drug candidates. Here, we present results from the application of a gated recurrent unit cell (GRU)-RNN to transfer learning on data sets of varying sizes and complexity. Analysis of the results has allowed us to provide some general guidelines for transfer learning. In particular, we show that data set sizes containing at least 190 molecules are needed for effective GRU-RNN-based molecular generation using transfer learning. The methods presented here should be applicable generally to the benchmarking of other deep learning methodologies for molecule generation.

Training Neural Nets To Learn Reactive Potential Energy Surfaces Using Interactive Quantum Chemistry in Virtual Reality.

While the primary bottleneck to a number of computational workflows was not so long ago limited by processing power, the rise of machine learning technologies has resulted in an interesting paradigm shift, which places increasing value on issues related to data curation-that is, data size, quality, bias, format, and coverage. Increasingly, data-related issues are equally as important as the algorithmic methods used to process and learn from the data. Here we introduce an open-source graphics processing unit-accelerated neural network (NN) framework for learning reactive potential energy surfaces (PESs). To obtain training data for this NN framework, we investigate the use of real-time interactive ab initio molecular dynamics in virtual reality (iMD-VR) as a new data curation strategy that enables human users to rapidly sample geometries along reaction pathways. Focusing on hydrogen abstraction reactions of CN radical with isopentane, we compare the performance of NNs trained using iMD-VR data versus NNs trained using a more traditional method, namely, molecular dynamics (MD) constrained to sample a predefined grid of points along the hydrogen abstraction reaction coordinate. Both the NN trained using iMD-VR data and the NN trained using the constrained MD data reproduce important qualitative features of the reactive PESs, such as a low and early barrier to abstraction. Quantitative analysis shows that NN learning is sensitive to the data set used for training. Our results show that user-sampled structures obtained with the quantum chemical iMD-VR machinery enable excellent sampling in the vicinity of the minimum energy path (MEP). As a result, the NN trained on the iMD-VR data does very well predicting energies that are close to the MEP but less well predicting energies for "off-path" structures. The NN trained on the constrained MD data does better predicting high-energy off-path structures, given that it included a number of such structures in its training set.

Adaptively Accelerating Reactive Molecular Dynamics Using Boxed Molecular Dynamics in Energy Space.

The problem of observing rare events is pervasive among the molecular dynamics community and an array of different types of methods are commonly used to accelerate these long time scale processes. Typically, rare event acceleration methods require an a priori specification of the event to be accelerated. In recent work, we have demonstrated the application of boxed molecular dynamics to energy space, as a way to accelerate rare events in the stochastic chemical master equation. Here we build upon this work and apply the boxed molecular dynamics algorithm to the energy space of a molecule in classical trajectory simulations. Through this new BXD in energy (BXDE) approach we demonstrate that generic rare events (in this case chemical reactions) may be accelerated by multiple orders of magnitude compared to unbiased simulations. Furthermore, we show that the ratios of products formed from the BXDE simulations are similar to those formed in unbiased simulations at the same temperature.

DFT Analysis of Spin Crossover in Mn(III) Complexes: Is a Two-Electron S = 2 to S = 0 Spin Transition Feasible?

Six-coordinate, rigorously octahedral d4 Mn(III) spin crossover (SCO) complexes are limited by symmetry to an S = 1 (intermediate spin, IS) to S = 2 (high spin, HS) transition. In order to realize the potential S = 0 to S = 2 transition, a lower symmetry and/or change in coordination number is needed, which we explore here computationally. First, a number of complexes are analyzed to develop a reliable and relatively fast DFT protocol for reproducing known Mn(III) spin state energetics. The hybrid meta-GGA functional TPSSh with a modest split valence plus polarization basis set and an empirical dispersion correction is found to predict correctly the ground spin state of Mn(III) complexes, including true low-spin (LS) S = 0 systems, with a range of donor sets including the hexadentate [N4O2] Schiff base ligands. The electronic structure design criteria necessary for realizing a ΔS = 2 SCO transition are described, and a number of model complexes are screened for potential SCO behavior. Five-coordinate trigonal-bipyramidal symmetry fails to yield any suitable systems. Seven-coordinate, approximately pentagonal bipyramidal symmetry is more favorable, and when a known pentadentate macrocyclic donor is combined with π-acceptor axial ligands, a novel Mn(III) complex, [Mn(PABODP)(PF3)2]3+ (PABODP = 2,13-dimethyl-3,6,9,12,18-pentaazabicyclo[12.3.1]octadeca-1(18),2,12,14,16-pentaene), is predicted to have the right spin state energetics for an S = 0 to S = 2 transition. Successful synthesis of such a complex could provide the first example of a ΔS = 2 SCO transition for d4 Mn(III). However, the combination of a rigid macrocycle and a high coordination number dilutes the stereochemical activity of the d electrons, leading to relatively small structural changes between HS and LS systems. It may therefore remain a challenge to realize strong cooperative effects in Mn(III) systems.