Deep generative model for drug design from protein target sequence.

Area of Science:

• Computational chemistry
• Bioinformatics
• Drug discovery

Background:

• Traditional drug discovery is time-consuming and expensive.
• Deep learning (DL) models accelerate drug discovery but often require prior molecular knowledge or protein binding site information.
• Existing DL methods for drug discovery rely heavily on known molecular structures or protein pocket data.

Discussion:

• DeepTarget is an end-to-end deep learning model designed for novel molecule generation.
• It utilizes a protein's amino acid sequence as the sole input, minimizing the need for prior knowledge.
• The model comprises three modules: Amino Acid Sequence Embedding (AASE), Structural Feature Inference (SFI), and Molecule Generation (MG).

Key Insights:

• DeepTarget successfully generates novel molecules conditioned solely on amino acid sequences.
• The generated molecules' validity was confirmed using a benchmark platform for molecular generation models.
• Drug-target affinity and molecular docking validated the interaction between generated molecules and target proteins.

Outlook:

• This approach significantly reduces the dependency on extensive prior knowledge in drug discovery.
• DeepTarget offers a promising direction for accelerating the identification of new drug candidates.
• Further research can explore integrating additional biological data to refine molecule generation.