Dr.Emb Appyter: A web platform for drug discovery using embedding vectors

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Embedding methods place similar compounds near each other in latent space, facilitating property-based searches.
• Traditional embedding methods often demand significant computational resources and programming proficiency.

Purpose of the Study:

• To develop a user-friendly, web-based platform, Dr.Emb Appyter, for chemical compound searching in drug discovery.
• To eliminate technical barriers associated with using embedding vectors for compound similarity analysis.

Main Methods:

• Utilizes various embedding techniques including fingerprinting, SMILES, and transcriptional response-based methods.
• Employs a Faiss-based search system for efficient retrieval of similar compounds.
• Integrates data visualization tools (3D scatter plots, heatmaps, UpSet plots) and drug-set enrichment analysis.

Main Results:

• Dr.Emb Appyter successfully embeds numerous compounds using diverse methods.
• The platform efficiently identifies compounds with similar properties based on embedding vector proximity.
• Provides comprehensive analysis and visualization of search results for drug discovery insights.

Conclusions:

• Dr.Emb Appyter democratizes the use of advanced embedding techniques for drug discovery.
• The platform offers a powerful, accessible tool for identifying novel drug candidates.
• Facilitates efficient and insightful compound similarity searches through a web-based interface.