Predicting fragment binding modes using customized Lennard-Jones potentials in short molecular dynamics simulations.

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Accurate prediction of fragment binding modes is essential for rational drug design.
• Fragments, due to their small size and low affinity, can exhibit diverse binding behaviors.
• Current in silico methods face challenges in reliably predicting these interactions.

Purpose of the Study:

• To develop and validate a computational workflow for predicting fragment binding sites and poses.
• To address the challenge of low binding affinity and diverse interactions of fragments.
• To provide a tool for fragment-based drug discovery, particularly when experimental data is limited.

Main Methods:

• Utilized multiple short molecular dynamics simulations to explore binding possibilities.
• Employed tailored Lennard-Jones potentials for simulating high fragment concentrations.
• Integrated descriptor analysis and binding free energy calculations for filtering and validation.

Main Results:

• The proposed workflow demonstrated high accuracy in identifying correct fragment binding sites.
• Successfully predicted native binding modes for fragments across four epigenetic target proteins.
• Validated the method's performance on established fragment-protein systems.

Conclusions:

• The developed computational approach offers a reliable alternative for predicting fragment binding modes.
• This method enhances the capabilities of fragment-based drug discovery.
• It is particularly valuable in scenarios with limited experimental structural data.