Abstract:In this work, we propose MEDICO, a Multi-viEw Deep generative model for molecule generation, structural optimization, and the SARS-CoV-2 Inhibitor disCOvery. To the best of our knowledge, MEDICO is the first-of-this-kind graph generative model that can generate molecular graphs similar to the structure of targeted molecules, with a multi-view representation learning framework to sufficiently and adaptively learn comprehensive structural semantics from targeted molecular topology and geometry. We show that our MEDICO significantly outperforms the state-of-the-art methods in generating valid, unique, and novel molecules under benchmarking comparisons. In particular, we showcase the multi-view deep learning model enables us to generate not only the molecules structurally similar to the targeted molecules but also the molecules with desired chemical properties, demonstrating the strong capability of our model in exploring the chemical space deeply. Moreover, case study results on targeted molecule generation for the SARS-CoV-2 main protease (Mpro) show that by integrating molecule docking into our model as chemical priori, we successfully generate new small molecules with desired drug-like properties for the Mpro, potentially accelerating the de novo design of Covid-19 drugs. Further, we apply MEDICO to the structural optimization of three well-known Mpro inhibitors (N3, 11a, and GC376) and achieve ~88% improvement in their binding affinity to Mpro, demonstrating the application value of our model for the development of therapeutics for SARS-CoV-2 infection.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to accelerate the discovery of SARS - CoV - 2 inhibitors through a multi - view deep - learning model. Specifically: 1. **Molecule Generation and Structural Optimization**: The paper proposes an innovative deep - generative model named MEDICO (Multi - viEw Deep generative model for molecule generation, structural optimization, and the SARS - CoV - 2 Inhibitor disCOvery). This model aims to generate new molecules similar to the target molecule structure and optimize the chemical properties of these molecules, thereby accelerating new drug design. 2. **Combining Multi - view Representation Learning Framework**: MEDICO introduces the multi - view representation learning framework for the first time, which can fully and adaptively learn comprehensive structural semantics from the topological structure and geometric information of the target molecule. This enables the model to generate molecules that are not only structurally similar but also have the required chemical properties, demonstrating its powerful ability to explore the chemical space in - depth. 3. **Design of SARS - CoV - 2 Main Protease (Mpro) Inhibitors**: By integrating the molecular docking results as chemical prior information into the model, MEDICO successfully generates new small molecules with ideal drug properties, especially for the main protease (Mpro) of SARS - CoV - 2, thus accelerating the de novo design of Covid - 19 drugs. 4. **Improving the Binding Affinity of Known Inhibitors**: The paper also applies MEDICO to perform structural optimization on three known Mpro inhibitors (N3, 11a, and GC376), achieving an approximately 88% increase in binding affinity, further proving the application value of this model in the development of drugs for treating SARS - CoV - 2 infection. ### Key Technical Points - **Multi - view Representation Learning**: MEDICO combines the topological structure and geometric information of molecules. It processes 2D topological structures through a flow - based model and processes 3D geometric information through a message - passing neural network (MPNN), and finally generates a joint representation. - **Molecular Docking as Chemical Prior**: Utilize the results of molecular docking (such as binding energy) to provide guidance for the model, ensuring that the generated molecules have potential antiviral activity. - **Performance Evaluation**: Evaluate the quality of the generated molecules through indicators such as effectiveness, novelty, and uniqueness, and evaluate the inhibitory effect of the generated molecules on Mpro through molecular docking scores. In conclusion, MEDICO aims to address the shortcomings of traditional molecule - generation methods in terms of structural similarity and chemical property prediction, providing new tools and methods for accelerating the discovery of SARS - CoV - 2 inhibitors.

Multi-view deep learning based molecule design and structural optimization accelerates the SARS-CoV-2 inhibitor discovery

Structure-based Virtual Screening, in Silico Docking, ADME Properties Prediction and Molecular Dynamics Studies for the Identification of Potential Inhibitors Against SARS-CoV-2 Mpro.

CogMol: Target-Specific and Selective Drug Design for COVID-19 Using Deep Generative Models

Structure-based <i>de novo</i> drug design using 3D deep generative models

De Novo Design and Bioactivity Prediction of SARS-CoV-2 Main Protease Inhibitors Using ULMFit

3D-Scaffold: A Deep Learning Framework to Generate 3D Coordinates of Drug-like Molecules with Desired Scaffolds

A dual diffusion model enables 3D molecule generation and lead optimization based on target pockets

Benchmarking Deep Graph Generative Models for Optimizing New Drug Molecules for COVID-19

Designing novel inhibitor derivatives targeting SARS-CoV-2 Mpro enzyme: a deep learning and structure biology approach

Development of a Deep Learning Generative Neural Network for Computer-Aided Design of Potential SARS-Cov-2 Inhibitors

De novo design and bioactivity prediction of SARS-CoV-2 main protease inhibitors using recurrent neural network-based transfer learning

AI-Aided Design of Novel Targeted Covalent Inhibitors against SARS-CoV-2

De Novo design of potential inhibitors against SARS-CoV-2 Mpro

Accelerating Inhibitor Discovery With A Deep Generative Foundation Model: Validation for SARS-CoV-2 Drug Targets

Predicting drug-Protein interaction with deep learning framework for molecular graphs and sequences: Potential candidates against SAR-CoV-2

Artificial Intelligence-Guided De Novo Molecular Design Targeting COVID-19

Multimolecular feature-based machine learning: system biology enhanced RF-QSAR modeling for the efficient prediction of the inhibitory potential of diverse SARS CoV-2 3CL Protease inhibitors

Molecule generation toward target protein (SARS-CoV-2) using reinforcement learning-based graph neural network via knowledge graph

Active learning driven prioritisation of compounds from on-demand libraries targeting the SARS-CoV-2 main protease

Multi-Objective De Novo Drug Design with Conditional Graph Generative Model

Scaffold-Constrained Molecular Generation