Ab initio quantum chemistry for protein structures.

Area of Science:

• Computational chemistry
• Structural biology
• Biophysics

Background:

• Experimental methods like X-ray crystallography and NMR provide reference protein structures.
• Ab initio force fields aim to replicate these experimental structures computationally.
• Assessing the accuracy of computational methods is crucial for structural biology.

Purpose of the Study:

• To evaluate the accuracy of ab initio force fields in predicting protein structural properties.
• To compare density-based and wave-function-based electronic structure methods for protein optimization.
• To determine the impact of basis set size on the quality of ab initio protein structures.

Main Methods:

• Utilized GPU-accelerated quantum chemistry algorithms (TeraChem) for ab initio protein structure optimization.
• Calculated structural properties for over 55 small proteins.
• Compared ab initio optimized structures against experimental (crystallography, NMR) and molecular mechanics references.

Main Results:

• The quality of ab initio optimized protein structures improves with increasing basis set size.
• Density functional theory (DFT) showed limited improvement over Hartree-Fock (HF) methods for predicted structures.
• Minimal basis sets sometimes caused structural issues, but these were resolved with double-ζ basis sets.

Conclusions:

• Ab initio force fields, particularly with larger basis sets, can reliably reproduce experimental protein structures.
• Basis set size is a key factor for accuracy in ab initio protein structure prediction.
• While DFT offers advantages in some contexts, HF methods with appropriate basis sets are effective for protein structure optimization.