Recent Advances in AI-Driven Drug Development: An Overview of MolProphet Platform

The integration of artificial intelligence (AI) has revolutionized the drug development landscape, accelerating the discovery and refinement of new therapeutics. Among these innovative technologies, MolProphet stands out as a comprehensive AI-based platform designed to enhance drug discovery processes.

Introduction of MolProphet Platform

MolProphet is an online collaborative platform aimed at making AI technology accessible to researchers working in drug discovery. It offers an all-in-one drug discovery platform that can be accessed online or privately, catering to the needs of investigators at various stages of new drug development, particularly in the early stages of hits discovery and hits-to-leads.

Main Functions and Modules of MolProphet

The platform primarily focuses on the early stages of drug discovery and is designed to be free for basic functionalities. Its core consists of a management system and six main functional modules that provide essential tools for researchers. These modules include:

Pocket Prediction and Definition Module

This module helps researchers define optimal target pocket structures, which significantly improve development efficiency. Utilizing deep learning algorithms, MolProphet models known protein pocket structures, offering multiple competitive pocket structures with a 3D visual selection interface. Future updates will enhance precision to the amino acid site level.

Structure-based Drug Discovery Module

To reduce research and development (RD) costs, this module leverages AI technology to perform virtual screening of commercial libraries. A deep neural network trained on 18 million Bioassay real data effectively predicts the affinity of unknown targets and ligands, achieving a billion-order molecule screening within 7 days. Future iterations will analyze activity data from earlier molecular cell experiments to refine screening.

Ligand-based Drug Discovery Module

This module utilizes AI algorithms to retrieve similar molecules from known commercial libraries. It includes a 2D structure similarity algorithm and 3D pharmacophore similarity algorithm, enabling the rapid retrieval of billion molecules in a minute and a day, respectively.

AI Molecular Design Module

This module employs AI to de novo design molecules, focusing on synthesizability and chemical synthesis pathways. Using 1000 reaction templates based on over 3 million explicit examples and expert-curated templates, it provides detailed synthesis solutions for each output molecule.

AI Docking Module

This module evaluates molecular datasets through AI, supporting rapid evaluation of ligand binding conformation using geometric deep learning and reinforcement learning for optimal binding conformation prediction.

Data Analysis Module

This module predicts interaction characteristics between molecules and ligands through multi-dimensional analysis, currently focusing on binding site analysis, binding force analysis, and binding conformation analysis. Additional analysis tools such as toxic sub-structure hints and proprietary molecular reports are in the works.

User Interface and Operations

MolProphet is designed to be user-friendly, requiring minimal software expertise. Each functional module is simplified into a few clicks, supplemented with necessary hints. The platform includes preset data submission methods, such as uploading local files or searching for PDB files from the official PDB website.

For result management, MolProphet offers task objectives and conclusion evaluation functions, facilitating easy review and communication. Users can collect, download, and delete results as needed. Detailed result analysis interfaces are provided for each specific result molecule, accessible via corresponding molecule tabs.

Project and Task Management

MolProphet establishes a project-based data management solution, allowing users to create multiple isolated projects and collaborate with different users within projects. An independent three-level management system, including PI, administrator, and general members, supports efficient team management.

The platform also offers task management features, enabling users to monitor progress, cancel, and retry tasks if needed. A hybrid cloud system is available for highly private scenarios, ensuring data security.

Conclusion

In summary, MolProphet provides users with a simple and comprehensive AI-driven drug discovery platform, currently accessible publicly. Future additions will include advanced services like lead optimization, further enhancing the platform's capabilities in the drug development process.