A backbone-dependent rotamer library with high (ϕ, ψ) coverage using metadynamics simulations

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Molecular Modeling

Background:

• Rotamer libraries are crucial for protein structure modeling, but existing ones struggle with constrained peptides due to limited backbone conformation data.
• Linear protein crystal structures do not fully represent the conformational space sampled by cyclic peptides.

Purpose of the Study:

• To develop a novel backbone-dependent rotamer library with high coverage across (ϕ, ψ) space.
• To improve the modeling of side chain dihedral angles for cyclic peptides and noncanonical amino acids.

Main Methods:

• Utilized explicit-solvent metadynamics simulations of model dipeptides to generate rotamer data.
• Created the Metadynamics of Dipeptides for Rotamer Distribution (MEDFORD) rotamer library.
• Validated the library against high-resolution protein crystal structures and available cyclic peptide data.

Main Results:

• The MEDFORD library demonstrates comparable accuracy (~80%) to state-of-the-art libraries for standard protein structures.
• MEDFORD shows distinct predictions in noncanonical (ϕ, ψ) regions compared to existing libraries.
• For noncanonical amino acids, MEDFORD achieves approximately 75% accuracy in predicting χ1 values.

Conclusions:

• The MEDFORD rotamer library offers improved modeling capabilities for peptides with noncanonical backbone conformations.
• This simulation-based approach is applicable for creating rotamer libraries for diverse amino acid types.
• Further data on cyclic peptides will be essential for comprehensive validation in these specific structures.