DFRscore: Deep Learning-Based Scoring of Synthetic Complexity with Drug-Focused Retrosynthetic Analysis for High-Throughput Virtual Screening.

Recently emerging generative AI models enable us to produce a vast number of compounds for potential applications. While they can provide novel molecular structures, the synthetic feasibility of the generated molecules is often questioned. To address this issue, a few recent studies have attempted to use deep learning models to estimate the synthetic accessibility of many molecules rapidly. However, retrosynthetic analysis tools used to train the models rely on reaction templates automatically extracted from a large reaction database that are not domain-specific and may exhibit low chemical correctness. To overcome this limitation, we introduce DFRscore (Drug-Focused Retrosynthetic score), a deep learning-based approach for a more practical assessment of synthetic accessibility in drug discovery. The DFRscore model is trained exclusively on drug-focused reactions, providing a predicted number of minimally required synthetic steps for each compound. This approach enables practitioners to filter out compounds that do not meet their desired level of synthetic accessibility at an early stage of high-throughput virtual screening for accelerated drug discovery. The proposed strategy can be easily adapted to other domains by adjusting the synthesis planning setup of the reaction templates and starting materials.

Molecular Generative Model via Retrosynthetically Prepared Chemical Building Block Assembly.

Deep generative models are attracting attention as a smart molecular design strategy. However, previous models often render molecules with low synthesizability, hindering their real-world applications. Here, a novel graph-based conditional generative model which makes molecules by tailoring retrosynthetically prepared chemical building blocks until achieving target properties in an auto-regressive fashion is proposed. This strategy improves the synthesizability and property control of the resulting molecules and also helps learn how to select appropriate building blocks and bind them together to achieve target properties. By applying a negative sampling method to the selection process of building blocks, this model overcame a critical limitation of previous fragment-based models, which can only use molecules from the training set during generation. As a result, the model works equally well with unseen building blocks without sacrificing computational efficiency. It is demonstrated that the model can generate potential inhibitors with high docking scores against the 3CL protease of SARS-COV-2.

PIGNet: a physics-informed deep learning model toward generalized drug-target interaction predictions.

Recently, deep neural network (DNN)-based drug-target interaction (DTI) models were highlighted for their high accuracy with affordable computational costs. Yet, the models' insufficient generalization remains a challenging problem in the practice of in silico drug discovery. We propose two key strategies to enhance generalization in the DTI model. The first is to predict the atom-atom pairwise interactions via physics-informed equations parameterized with neural networks and provides the total binding affinity of a protein-ligand complex as their sum. We further improved the model generalization by augmenting a broader range of binding poses and ligands to training data. We validated our model, PIGNet, in the comparative assessment of scoring functions (CASF) 2016, demonstrating the outperforming docking and screening powers than previous methods. Our physics-informing strategy also enables the interpretation of predicted affinities by visualizing the contribution of ligand substructures, providing insights for further ligand optimization.

Drug-likeness scoring based on unsupervised learning.

Drug-likeness prediction is important for the virtual screening of drug candidates. It is challenging because the drug-likeness is presumably associated with the whole set of necessary properties to pass through clinical trials, and thus no definite data for regression is available. Recently, binary classification models based on graph neural networks have been proposed but with strong dependency of their performances on the choice of the negative set for training. Here we propose a novel unsupervised learning model that requires only known drugs for training. We adopted a language model based on a recurrent neural network for unsupervised learning. It showed relatively consistent performance across different datasets, unlike such classification models. In addition, the unsupervised learning model provides drug-likeness scores that well separate distributions with increasing mean values in the order of datasets composed of molecules at a later step in a drug development process, whereas the classification model predicted a polarized distribution with two extreme values for all datasets presumably due to the overconfident prediction for unseen data. Thus, this new concept offers a pragmatic tool for drug-likeness scoring and further can be applied to other biochemical applications.

ACE-Molecule: An open-source real-space quantum chemistry package.

ACE-Molecule (advanced computational engine for molecules) is a real-space quantum chemistry package for both periodic and non-periodic systems. ACE-Molecule adopts a uniform real-space numerical grid supported by the Lagrange-sinc functions. ACE-Molecule provides density functional theory (DFT) as a basic feature. ACE-Molecule is specialized in efficient hybrid DFT and wave-function theory calculations based on Kohn-Sham orbitals obtained from a strictly localized exact exchange potential. It is open-source oriented calculations with a flexible and convenient development interface. Thus, ACE-Molecule can be improved by actively adopting new features from other open-source projects and offers a useful platform for potential developers and users. In this work, we introduce overall features, including theoretical backgrounds and numerical examples implemented in ACE-Molecule.

Homochiral Supramolecular Thin Film from Self-Assembly of Achiral Triarylamine Molecules by Circularly Polarized Light.

Here, we report the formation of homochiral supramolecular thin film from achiral molecules, by using circularly polarized light (CPL) only as a chiral source, on the condition that irradiation of CPL does not induce a photochemical change of the achiral molecules. Thin films of self-assembled structures consisting of chiral supramolecular fibrils was obtained from the triarylamine derivatives through evaporation of the self-assembled triarylamine solution. The homochiral supramolecular helices with the desired handedness was achieved by irradiation of circularly polarized visible light during the self-assembly process, and the chiral stability of supramolecular self-assembled product was achieved by photopolymerization of the diacetylene moieties at side chains of the building blocks, with irradiation of circularly polarized ultraviolet light. This work provides a novel methodology for the generation of homochiral supramolecular thin film from the corresponding achiral molecules.

Molecular Generative Model Based on an Adversarially Regularized Autoencoder.

Deep generative models are attracting great attention as a new promising approach for molecular design. A variety of models reported so far are based on either a variational autoencoder (VAE) or a generative adversarial network (GAN), but they have limitations such as low validity and uniqueness. Here, we propose a new type of model based on an adversarially regularized autoencoder (ARAE). It basically uses latent variables like VAE, but the distribution of the latent variables is estimated by adversarial training like in GAN. The latter is intended to avoid both the insufficiently flexible approximation of posterior distribution in VAE and the difficulty in handling discrete variables in GAN. Our benchmark study showed that ARAE indeed outperformed conventional models in terms of validity, uniqueness, and novelty per generated molecule. We also demonstrated a successful conditional generation of drug-like molecules with ARAE for the control of both cases of single and multiple properties. As a potential real-world application, we could generate epidermal growth factor receptor inhibitors sharing the scaffolds of known active molecules while satisfying drug-like conditions simultaneously.

Predicting Drug-Target Interaction Using a Novel Graph Neural Network with 3D Structure-Embedded Graph Representation.

We propose a novel deep learning approach for predicting drug-target interaction using a graph neural network. We introduce a distance-aware graph attention algorithm to differentiate various types of intermolecular interactions. Furthermore, we extract the graph feature of intermolecular interactions directly from the 3D structural information on the protein-ligand binding pose. Thus, the model can learn key features for accurate predictions of drug-target interaction rather than just memorize certain patterns of ligand molecules. As a result, our model shows better performance than docking and other deep learning methods for both virtual screening (AUROC of 0.968 for the DUD-E test set) and pose prediction (AUROC of 0.935 for the PDBbind test set). In addition, it can reproduce the natural population distribution of active molecules and inactive molecules.

Scaffold-based molecular design with a graph generative model.

Searching for new molecules in areas like drug discovery often starts from the core structures of known molecules. Such a method has called for a strategy of designing derivative compounds retaining a particular scaffold as a substructure. On this account, our present work proposes a graph generative model that targets its use in scaffold-based molecular design. Our model accepts a molecular scaffold as input and extends it by sequentially adding atoms and bonds. The generated molecules are then guaranteed to contain the scaffold with certainty, and their properties can be controlled by conditioning the generation process on desired properties. The learned rule of extending molecules can well generalize to arbitrary kinds of scaffolds, including those unseen during learning. In the conditional generation of molecules, our model can simultaneously control multiple chemical properties despite the search space constrained by fixing the substructure. As a demonstration, we applied our model to designing inhibitors of the epidermal growth factor receptor and show that our model can employ a simple semi-supervised extension to broaden its applicability to situations where only a small amount of data is available.

Study of Li Adsorption on Graphdiyne Using Hybrid DFT Calculations.

Promising applications of graphdiyne have often been initiated by theoretical predictions especially using DFT known as the most powerful first-principles electronic structure calculation method. However, there is no systematic study on the reliability of DFT for the prediction of the electronic properties of the graphdiyne. Here, we performed a study of Li adsorption on the graphdiyne using hybrid DFT with LC-ωPBE and compared the results with those of PBE, because accurate prediction of the Li adsorption is important for performance as a Li storage that was first theoretically suggested and then experimentally realized. Our results show that PBE overestimates the adsorption energy inside a pore and the barrier height at the transition state of in-plane diffusion compared to the those of LC-ωPBE. In particular, LC-ωPBE predicted almost barrier-less in-plane diffusion of Li on the graphdiyne because of the presence of both in-plane and out-of-plane π orbitals. Also, LC-ωPBE favors a high spin state due to the exact exchange energy when several Li atoms are adsorbed on the graphdiyne, whereas PBE favors a low spin state. Thus, the use of the hybrid DFT is critical for reliable predictions on the electronic properties of the graphdiyne.

Molecular generative model based on conditional variational autoencoder for de novo molecular design.

We propose a molecular generative model based on the conditional variational autoencoder for de novo molecular design. It is specialized to control multiple molecular properties simultaneously by imposing them on a latent space. As a proof of concept, we demonstrate that it can be used to generate drug-like molecules with five target properties. We were also able to adjust a single property without changing the others and to manipulate it beyond the range of the dataset.

Non-empirical atomistic dipole-interaction-model for quantum plasmon simulation of nanoparticles.

Plasmonic nanoparticles in the quantum regime exhibit characteristic optical properties that cannot be described by classical theories. Time-dependent density functional theory (TDDFT) is rising as a versatile tool for study on such systems, but its application has been limited to very small clusters due to rapidly growing computational costs. We propose an atomistic dipole-interaction-model for quantum plasmon simulations as a practical alternative. Namely the atomic dipole approximation represents induced dipoles with atomic polarizabilities obtained from TDDFT without empirical parameters. It showed very good agreement with TDDFT for plasmonic spectra of small silver clusters at much lower computational cost, though it is not appropriate for molecular-like excitations. It could also reproduce the plasmonic band shift experimentally observed in sub-10 nm silver particles.

Outstanding performance of configuration interaction singles and doubles using exact exchange Kohn-Sham orbitals in real-space numerical grid method.

To assess the performance of multi-configuration methods using exact exchange Kohn-Sham (KS) orbitals, we implemented configuration interaction singles and doubles (CISD) in a real-space numerical grid code. We obtained KS orbitals with the exchange-only optimized effective potential under the Krieger-Li-Iafrate (KLI) approximation. Thanks to the distinctive features of KLI orbitals against Hartree-Fock (HF), such as bound virtual orbitals with compact shapes and orbital energy gaps similar to excitation energies; KLI-CISD for small molecules shows much faster convergence as a function of simulation box size and active space (i.e., the number of virtual orbitals) than HF-CISD. The former also gives more accurate excitation energies with a few dominant configurations than the latter, even with many more configurations. The systematic control of basis set errors is straightforward in grid bases. Therefore, grid-based multi-configuration methods using exact exchange KS orbitals provide a promising new way to make accurate electronic structure calculations.