A gradient-directed Monte Carlo approach to molecular design

SimTune: bridging the simulator reality gap for resource management in edge-cloud computing

Industries and services are undergoing an Internet of Things centric transformation globally, giving rise to an explosion of multi-modal data generated each second. This, with the requirement of low-latency result delivery, has led to the ubiquitous adoption of edge and cloud computing paradigms. Edge computing follows the data gravity principle, wherein the computational devices move closer to the end-users to minimize data transfer and communication times. However, large-scale computation has exacerbated the problem of efficient resource management in hybrid edge-cloud platforms. In this regard, data-driven models such as deep neural networks (DNNs) have gained popularity to give rise to the notion of edge intelligence. However, DNNs face significant problems of data saturation when fed volatile data. Data saturation is when providing more data does not translate to improvements in performance. To address this issue, prior work has leveraged coupled simulators that, akin to digital twins, generate out-of-distribution training data alleviating the data-saturation problem. However, simulators face the reality-gap problem, which is the inaccuracy in the emulation of real computational infrastructure due to the abstractions in such simulators. To combat this, we develop a framework, SimTune, that tackles this challenge by leveraging a low-fidelity surrogate model of the high-fidelity simulator to update the parameters of the latter, so to increase the simulation accuracy. This further helps co-simulated methods to generalize to edge-cloud configurations for which human encoded parameters are not known apriori. Experiments comparing SimTune against state-of-the-art data-driven resource management solutions on a real edge-cloud platform demonstrate that simulator tuning can improve quality of service metrics such as energy consumption and response time by up to 14.7% and 7.6% respectively.

iQSPR in XenonPy: A Bayesian Molecular Design Algorithm

iQSPR is an inverse molecular design algorithm based on Bayesian inference that was developed in our previous study. Here, the algorithm is integrated in Python as a new module called iQSPR-X in the all-in-one materials informatics platform XenonPy. Our new software provides a flexible, easy-to-use, and extensible platform for users to build customized molecular design algorithms using pre-set modules and a pre-trained model library in XenonPy. In this paper, we describe key features of iQSPR-X and provide guidance on its use, illustrated by an application to a polymer design that targets a specific range of bandgap and dielectric constant.

Acceleration of Inverse Molecular Design by Using Predictive Techniques

This study addresses one of the most important drawbacks inherently related to molecular searches in chemical compound space by greedy algorithms such as Best First Search and Genetic Algorithm, i.e., the large computational cost required to optimize one or more quantum-chemical properties. Significant speed-ups are obtained by initial property screening via predictive techniques starting already from very small databases. It is shown that the attainable acceleration depends heavily on the molecular properties, the predictive model, the molecular descriptor, and the current size of the database. We discuss the implementation and performance of predictive techniques in molecular searches based on a fixed molecular framework with a selection of sites to be filled with groups from a chemical fragment library. It is shown that for some properties speed-ups of a factor of 5 to even 20 can be obtained, while inverse design procedures on more complex properties still reach speed-ups of a factor of 2 without losing performance.

Search for Catalysts by Inverse Design: Artificial Intelligence, Mountain Climbers, and Alchemists

In silico catalyst design is a grand challenge of chemistry. Traditional computational approaches have been limited by the need to compute properties for an intractably large number of possible catalysts. Recently, inverse design methods have emerged, starting from a desired property and optimizing a corresponding chemical structure. Techniques used for exploring chemical space include gradient-based optimization, alchemical transformations, and machine learning. Though the application of these methods to catalysis is in its early stages, further development will allow for robust computational catalyst design. This review provides an overview of the evolution of inverse design approaches and their relevance to catalysis. The strengths and limitations of existing techniques are highlighted, and suggestions for future research are provided.

Tuning the HOMO-LUMO Energy Gap of Small Diamondoids Using Inverse Molecular Design

Functionalized diamondoids show great potential as building blocks for various new optoelectronic applications. However, until now, only simple mono and double substitutions were investigated. In this work, we considered up to 10 and 6 sites for functionalization of the two smallest diamondoids, adamantane and diamantane, respectively, in search for diamondoid derivatives with a minimal and maximal HOMO-LUMO energy gap. To this end, the energy gap was optimized systematically using an inverse molecular design methodology based on the best-first search algorithm combined with a Monte Carlo component to escape local optima. Adamantane derivatives were found with HOMO-LUMO gaps ranging from 2.42 to 10.63 eV, with 9.45 eV being the energy gap of pure adamantane. For diamantane, similar values were obtained. The structures with the lowest HOMO-LUMO gaps showed apparent push-pull character. The push character is mainly formed by sulfur or nitrogen dopants and thiol groups, whereas the pull character is predominantly determined by the presence of electron-withdrawing nitro or carbonyl groups assisted by amino and hydroxyl groups via the formation of intramolecular hydrogen bonds. In contrast, maximal HOMO-LUMO gaps were obtained by introducing numerous electronegative groups.

Alchemical screening of ionic crystals

We introduce alchemical perturbations as a rapid and accurate tool to estimate fundamental structural and energetic properties in pure and mixed ionic crystals.

Efficient Computational Screening of Organic Polymer Photovoltaics

There has been increasing interest in rational, computationally driven design methods for materials, including organic photovoltaics (OPVs). Our approach focuses on a screening "pipeline", using a genetic algorithm for first stage screening and multiple filtering stages for further refinement. An important step forward is to expand our diversity of candidate compounds, including both synthetic and property-based measures of diversity. For example, top monomer pairs from our screening are all donor-donor (D-D) combinations, in contrast with the typical donor-acceptor (D-A) motif used in organic photovoltaics. We also find a strong "sequence effect", in which the average HOMO-LUMO gap of tetramers changes by ∼0.2 eV as a function of monomer sequence (e.g., ABBA versus BAAB); this has rarely been explored in conjugated polymers. Beyond such optoelectronic optimization, we discuss other properties needed for high-efficiency organic solar cells, and applications of screening methods to other areas, including non-fullerene n-type materials, tandem cells, and improving charge and exciton transport.

Stochastic voyages into uncharted chemical space produce a representative library of all possible drug-like compounds

The "small molecule universe" (SMU), the set of all synthetically feasible organic molecules of 500 Da molecular weight or less, is estimated to contain over 10(60) structures, making exhaustive searches for structures of interest impractical. Here, we describe the construction of a "representative universal library" spanning the SMU that samples the full extent of feasible small molecule chemistries. This library was generated using the newly developed Algorithm for Chemical Space Exploration with Stochastic Search (ACSESS). ACSESS makes two important contributions to chemical space exploration: it allows the systematic search of the unexplored regions of the small molecule universe, and it facilitates the mining of chemical libraries that do not yet exist, providing a near-infinite source of diverse novel compounds.

Liquid water simulations with the density fragment interaction approach

We reformulate the density fragment interaction (DFI) approach [Fujimoto and Yang, J. Chem. Phys., 2008, 129, 054102.] to achieve linear-scaling quantum mechanical calculations for large molecular systems. Two key approximations are developed to improve the efficiency of the DFI approach and thus enable the calculations for large molecules: the electrostatic interactions between fragments are computed efficiently by means of polarizable electrostatic-potential-fitted atomic charges; and frozen fragment pseudopotentials, similar to the effective fragment potentials that can be fitted from interactions between small molecules, are employed to take into account the Pauli repulsion effect among fragments. Our reformulated and parallelized DFI method demonstrates excellent parallel performance based on the benchmarks for the system of 256 water molecules. Molecular dynamics simulations for the structural properties of liquid water also show a qualitatively good agreement with experimental measurements including the heat capacity, binding energy per water molecule, and the radial distribution functions of atomic pairs of O-O, O-H, and H-H. With this approach, large-scale quantum mechanical simulations for water and other liquids become feasible.

A gradient-directed Monte Carlo approach for protein design

We develop a new global optimization strategy, gradient-directed Monte Carlo (GDMC) sampling, to optimize protein sequence for a target structure using RosettaDesign. GDMC significantly improves the sampling of sequence space, compared to the classical Monte Carlo search protocol, for a fixed backbone conformation as well as for the simultaneous optimization of sequence and structure. As such, GDMC sampling enhances the efficiency of protein design.

A gradient-directed Monte Carlo method for global optimization in a discrete space: application to protein sequence design and folding

We apply the gradient-directed Monte Carlo (GDMC) method to select optimal members of a discrete space, the space of chemically viable proteins described by a model Hamiltonian. In contrast to conventional Monte Carlo approaches, our GDMC method uses local property gradients with respect to chemical variables that have discrete values in the actual systems, e.g., residue types in a protein sequence. The local property gradients are obtained from the interpolation of discrete property values, following the linear combination of atomic potentials scheme developed recently [M. Wang et al., J. Am. Chem. Soc. 128, 3228 (2006)]. The local property derivative information directs the search toward the global minima while the Metropolis criterion incorporated in the method overcomes barriers between local minima. Using the simple HP lattice model, we apply the GDMC method to protein sequence design and folding. The GDMC algorithm proves to be particularly efficient, suggesting that this strategy can be extended to other discrete optimization problems in addition to inverse molecular design.

Discrete Optimization of Electronic Hyperpolarizabilities in a Chemical Subspace

We introduce a general optimization algorithm based on an interpolation of property values on a hypercube. Each vertex of the hypercube represents a molecule, while the interior of the interpolation represents a virtual superposition ("alchemical" mutation) of molecules. The resultant algorithm is similar to branch-and-bound/tree-search methods. We apply the algorithm to the optimization of the first electronic hyperpolarizability for several tolane libraries. The search includes structural and conformational information. Geometries were optimized using the AM1 Hamiltonian, and first hyperpolarizabilities were computed using the INDO/S method. Even for small libraries, a significant improvement of the hyperpolarizability, up to a factor of ca. 4, was achieved. The algorithm was validated for efficiency and reproduced known experimental results. The algorithm converges to a local optimum at a computational cost on the order of the logarithm of the library size, making large libraries accessible. For larger libraries, the improvement was accomplished by performing electronic structure calculations on less than 0.01% of the compounds in the larger libraries. Alternation of electron donating and accepting groups in the tolane scaffold was found to produce candidates with large hyperpolarizabilities consistently.

Exploring chemical space with discrete, gradient, and hybrid optimization methods

Discrete, gradient, and hybrid optimization methods are applied to the challenge of discovering molecules with optimized properties. The cost and performance of the approaches were studied using a tight-binding model to maximize the static first electronic hyperpolarizability of molecules. Our analysis shows that discrete branch and bound methods provide robust strategies for inverse chemical design involving diverse chemical structures. Based on the linear combination of atomic potentials, a hybrid discrete-gradient optimization strategy significantly improves the performance of the gradient methods. The hybrid method performs better than dead-end elimination and competes with branch and bound and genetic algorithms. The branch and bound methods for these model Hamiltonians are more cost effective than genetic algorithms for moderate-sized molecular optimization.

Computational design, synthesis and biological evaluation of para-quinone-based inhibitors for redox regulation of the dual-specificity phosphatase Cdc25B

Quinoid inhibitors of Cdc25B were designed based on the Linear Combination of Atomic Potentials (LCAP) methodology. In contrast to a published hypothesis, the biological activities and hydrogen peroxide generation in reducing media of three synthetic models did not correlate with the quinone half-wave potential, E(1/2).