Multiscale generalized born modeling of ligand binding energies for virtual database screening

Area of Science:

• Computational chemistry
• Molecular modeling
• Biophysics

Background:

• Generalized Born (GB) models are essential for studying electrostatic interactions in molecules like proteins.
• Accurate effective Born radii are crucial for satisfactory solvation energies.
• Previous studies indicated that GB models reproducing solvation energy may not be ideal for ligand binding calculations.

Purpose of the Study:

• To investigate ligand binding energetics using an exact Generalized Born (GB) model with Born radii derived from the Poisson-Boltzmann (PB) equation.
• To address the computational time constraints of accurate Born radii calculations in large-scale database screening.
• To develop a multiscale GB approach for efficient and accurate virtual screening.

Main Methods:

• Employed the exact GB model where Born radii are computed via the Poisson-Boltzmann (PB) equation.
• Introduced a multiscale GB approach dividing atoms into two groups.
• For critical atoms, accurate Born radii were computed; for others, fast GB methods or variational approaches were used.

Main Results:

• Accurate Born radii yielded excellent agreement between GB and PB electrostatic calculations for ligand binding.
• The proposed multiscale GB approach offers a balance between computational accuracy and speed.
• This strategy is suitable for the virtual screening of large molecular databases.

Conclusions:

• Accurate Born radii are vital for reliable GB model performance in ligand binding studies.
• The multiscale GB approach provides a computationally efficient alternative for large-scale screening.
• This method enhances accuracy while maintaining the speed required for practical applications in drug discovery.