In silico discovery of novel compounds for FAK activation using virtual screening, AI-based prediction, and molecular...

Area of Science:

• Biochemistry and Molecular Biology
• Computational Chemistry and Drug Discovery

Background:

• Focal Adhesion Kinase (FAK) is a critical non-receptor tyrosine kinase involved in cell signaling, proliferation, and migration.
• FAK overexpression is linked to metastatic and advanced-stage cancers, but its activity can decrease in other diseases.
• There is a need for compounds that can modulate FAK activity, particularly to enhance it in disease contexts.

Purpose of the Study:

• To identify novel compounds capable of enhancing Focal Adhesion Kinase (FAK) activity.
• To leverage structure-based virtual screening and artificial intelligence (AI) for drug discovery.
• To screen a large chemical database for potential FAK activity enhancers.

Main Methods:

• Utilized a structure-based virtual screening pipeline on a database of over 10 million compounds.
• Employed Tanimoto Similarity to identify compounds structurally related to a known FAK activator (ZINC40099027).
• Applied K-means clustering, molecular docking, deep learning (GLAM for BBB permeability, elEmBERT for toxicity), SAScorer, and 50 ns Molecular Dynamics (MD) simulations for compound evaluation.

Main Results:

• Screened over 10 million compounds, identifying 10 promising candidates based on similarity, docking, AI predictions, and physicochemical properties.
• Conducted MD simulations to assess the stability of the top 10 compounds' interaction with FAK.
• Identified the top three most promising candidate compounds after rigorous in silico evaluation.

Conclusions:

• The study successfully identified three novel compounds with the potential to enhance FAK activity.
• The integrated approach combining virtual screening, AI, and molecular dynamics is effective for discovering potential drug candidates.
• These identified compounds warrant further experimental validation for therapeutic development.