Backmapping of Highly Coarse-Grained Protein Models and Applications to Nanomedicine Designs

Area of Science:

• Computational Chemistry
• Biophysics
• Structural Biology

Background:

• Multiscale molecular modeling is crucial for studying complex biological assemblies.
• Reconstructing all-atom (AA) structures from coarse-grained (CG) models is a significant challenge.
• Current methods limit atomic detail retrieval at mesoscopic scales.

Purpose of the Study:

• To introduce a progressive backmapping strategy for reconstructing AA models from highly coarse-grained (HCG) models.
• To enable accurate atomic detail incorporation into mesoscale simulations.
• To support hierarchical reconstruction of complex protein assemblies.

Main Methods:

• Developed a progressive backmapping strategy within a hierarchical modeling framework.
• Utilized a neural-network-based approach (ProNet Backmapping) for gradual resolution refinement.
• Applied the method to various proteins, flexible linkers, and virus-like particles.

Main Results:

• Achieved accurate AA structure reconstructions across diverse proteins and flexible linkers.
• Successfully performed hierarchical backmapping of complex protein assemblies, including large virus-like particles.
• Demonstrated reconstruction from HCG to full AA resolution across multiple levels, including chemical variations like mutations.

Conclusions:

• The progressive backmapping strategy provides a scalable and rigorous framework for incorporating atomistic detail into mesoscale simulations.
• This methodology overcomes limitations in reconstructing atomic details from CG models.
• The approach is applicable to complex systems in chemistry and nanomedicine design.