Multi-eGO: Model Improvements toward the Study of Complex Self-Assembly Processes.

Area of Science:

• Computational Biology
• Biophysics
• Protein Dynamics

Background:

• Structure-based models are crucial for simulating protein folding mechanisms.
• Classical molecular dynamics can simulate fast-folding proteins but struggles with complex self-assembly.
• Protein aggregation and other out-of-equilibrium processes remain computationally challenging.

Purpose of the Study:

• To improve the multi-eGO hybrid model for simulating protein self-assembly.
• To demonstrate the model's capability in learning conformational ensembles and reproducing known mechanisms.
• To extend simulation capabilities to concentration-dependent and out-of-equilibrium processes.

Main Methods:

• Development and application of an improved hybrid multistate structure-based model, multi-eGO.
• Learning the conformational ensemble of amyloid beta 42 (Aβ42) intrinsically disordered peptide.
• Simulating the folding mechanism of the B1 immunoglobulin-binding domain.
• Reproducing the concentration-dependent aggregation of transthyretin (TTR) 105-115 peptide.

Main Results:

• The multi-eGO model accurately learned the conformational ensemble of Aβ42.
• The model successfully reproduced the folding mechanism of the B1 immunoglobulin-binding domain.
• The model accurately reproduced the aggregation of the TTR 105-115 peptide as a function of concentration.

Conclusions:

• The enhanced multi-eGO model effectively simulates diverse protein dynamics, including folding and aggregation.
• This model offers a promising platform for simulating complex self-assembly processes currently inaccessible to other techniques.
• Learning from minimal conformational states enables accurate prediction of protein self-assembly dynamics.