Systematic Partitioning of Proteins for Quantum-Chemical Fragmentation Methods Using Graph Algorithms.

Area of Science:

• Computational chemistry
• Biophysics
• Quantum chemistry

Background:

• Quantum-chemical fragmentation methods enable efficient analysis of large biomolecules.
• Accurate calculation of local properties in proteins depends heavily on fragmentation strategy.
• Current fixed-size fragmentation methods can introduce significant, unpredictable errors.

Purpose of the Study:

• To develop a systematic partitioning scheme for quantum-chemical fragmentation that minimizes errors for local target quantities.
• To improve the accuracy of protein property calculations by optimizing fragment selection.

Main Methods:

• Representing proteins as weighted graphs where amino acids are nodes and edge weights estimate fragmentation error.
• Employing graph partitioning algorithms to find near-optimal protein partitions.
• Applying a simplified molecular fractionation with conjugate caps (MFCC) approach with hydrogen caps.

Main Results:

• The graph-based partitioning scheme consistently reduces fragmentation errors compared to the naive approach.
• Demonstrated improved accuracy across six diverse protein test cases.
• Validated the scheme's effectiveness for various applications of fragmentation methods.

Conclusions:

• The developed graph-based scheme offers a more reliable and accurate method for fragmenting proteins in quantum-chemical calculations.
• This systematic approach enhances the predictability and precision of local property calculations in large biomolecules.
• The method provides a significant advancement over conventional fragmentation techniques.