Accuracy issues involved in modeling in vivo protein structures using PM7.

Area of Science:

• Computational Chemistry
• Structural Biology
• Biophysics

Background:

• Accurate prediction of in vivo protein structures is crucial for understanding biological function.
• Semiempirical quantum mechanical methods, like PM7, are used to model protein structures.
• X-ray crystallography provides high-resolution structural data but has inherent limitations.

Purpose of the Study:

• To quantify the error in predicting in vivo protein structures using the PM7 method.
• To identify and analyze the contributions of X-ray crystallography limitations, crystal vs. solution environments, and PM7 method errors.

Main Methods:

• Optimization of 19 high-accuracy protein structures from the Protein Data Bank using the PM7 semiempirical method.
• Calculation of the drop in heat of formation for optimized protein geometries.
• Validation of PM7-generated geometries using the Molprobity program to assess Clashscores.

Main Results:

• PM7 method errors account for approximately 10% of the calculated heat of formation decrease.
• X-ray structure limitations and differences between crystal and solution environments contribute the remaining error.
• A novel fault in PM7 was identified, involving underestimation of core-core repulsion, leading to unrealistic atomic contacts.

Conclusions:

• The PM7 method has limitations that affect protein structure prediction accuracy.
• Understanding these limitations is essential for reliable computational modeling of protein structures.
• Further refinement of semiempirical methods is needed to improve the prediction of protein conformations.