Scaffold-Retained Structure Generator to Exhaustively Create Molecules in an Arbitrary Chemical Space

Abstract

The construction of a virtual library (VL) consisting of novel molecules based on structure–activity relationships is crucial for lead optimization in rational drug design. In this study, we propose a novel scaffold-retained structure generator, EMPIRE (Exhaustive Molecular library Production In a scaffold-REtained manner), to create novel molecules in an arbitrary chemical space. By combining a deep learning model-based generator and a building block-based generator, the proposed method efficiently provides a VL consisting of molecules that retain the input scaffold and contain unique arbitrary substructures. The proposed method enables us to construct rational VLs located in unexplored chemical spaces containing molecules with unique skeletons (e.g., bicyclo[1.1.1]­pentane and cubane) or elements (e.g., boron and silicon). We expect EMPIRE to contribute to efficient drug design with unique substructures by virtual screening.

Application of Machine Learning to the Design of Energetic Materials: Preliminary Experience and Comparison with Alternative Techniques

“…Briefly, deep generative models broadly fall into four categories: Recursive Neural Networks (RNN) [59][60][61][62][63], Generative Adversarial Networks (GAN) [64][65][66], Variational Auto-Encoders (VAE) [67][68][69][70][71][72][73][74][75][76][77] and reinforcement learning (RL)-based strategies [78][79][80][81][82][83]. The inception technique is also noteworthy for its simplicity [84].…”

COMA: efficient structure-constrained molecular generation using contractive and margin losses

“…Modof is a graph-based molecular generation method that identifies editable sites in molecules using the graph edit distance algorithm and applies several graph edit methods, such as removal fragment prediction and child node type prediction, to improve a target property [ 5 ]. EMPIRE is a SMILES-based molecular editing method to generate a new molecule with desired scaffold structures [ 15 ]. EMPIRE first identifies the scaffold structure of an input molecule, generate molecular fragments via VAE-based and building block-based models, and produce new molecular structures that resemble the input scaffold structures by adding the generate fragments to the scaffold.…”

Fragment‐based deep molecular generation using hierarchical chemical graph representation and multi‐resolution graph variational autoencoder

“…Their model also incorporated reinforcement learning to guide the decorative process and optimize molecular properties. Kaitoh et al [17] proposed a new hybrid scaffold-constrained method named EMPIRE that combines the deep learning-based molecular generator with the traditional fragment enumeration method. Fragments generated by the VAE and the enumeration models are merged and subsequently attached to the input scaffold to form new molecules in an arbitrary chemical subspace.…”

“…Several SMILES-based models [15][16][17] focused on scaffold-constrained or scaffold-retained molecular generation. Josep Arús-Pous et al [15].…”