How well do semiempirical QM methods describe the structure of proteins?

Area of Science:

• Computational chemistry
• Structural biology
• Biophysics

Background:

• Semiempirical quantum-mechanical (QM) methods are widely used for biomolecular simulations.
• These methods were primarily developed and validated using small molecules.
• Their accuracy in describing larger biomolecular structures like proteins remains an active area of research.

Purpose of the Study:

• To evaluate the capability of various semiempirical QM methods in describing protein structures.
• To compare the performance of these QM methods against established molecular mechanics force fields.

Main Methods:

• Analysis of 19 high-resolution protein crystal structures.
• Geometry optimization of protein structures using semiempirical QM methods: PM6-D3H4, PM7, and GFN2-xTB.
• Comparison of QM-optimized structures with experimental crystal structures and AMBER ff03 force field results.

Main Results:

• Some semiempirical QM methods accurately capture general protein structural features.
• Performance of tested QM methods is comparable to or better than the AMBER ff03 force field.
• PM7 and PM6-D3H4 methods introduced artificial close contacts in protein structures.
• Reparameterization partially resolved the issue of artificial close contacts in PM7 and PM6-D3H4.

Conclusions:

• Semiempirical QM methods show promise for protein structure modeling in biomolecular studies.
• Careful selection and potential reparameterization of QM methods are crucial to avoid artifacts like artificial close contacts.
• GFN2-xTB and other suitable QM methods can be reliable tools for protein structure analysis.