Exploring CPS-Extrapolated DLPNO-CCSD(T₁) Reference Values for Benchmarking DFT Methods on Enzymatically Catalyzed...

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Method Development

Background:

• Domain-based local pair natural orbital coupled-cluster singles doubles with perturbative triples [DLPNO-CCSD(T)] offers a computationally efficient route to high-accuracy reference values.
• Extrapolation to the complete PNO space (CPS) has been proposed to further enhance the accuracy of DLPNO-CCSD(T) calculations.
• Density Functional Approximations (DFAs) require reliable benchmark data for validation, particularly for complex systems like enzyme active sites.

Purpose of the Study:

• To evaluate the impact of two complete PNO space (CPS) extrapolation levels, CPS(5,6) and CPS(6,7), on the accuracy of DLPNO-CCSD(T) benchmark values.
• To assess how these enhanced benchmark values affect the performance evaluation of Density Functional Approximations (DFAs) for organic and metalloenzyme active site models.
• To determine the suitability of CPS extrapolation as a cost-effective method for improving DLPNO-CCSD(T) accuracy in benchmarking.

Main Methods:

• Application of DLPNO-CCSD(T) calculations with CPS(5,6) and CPS(6,7) extrapolations to organic and transition-metal-dependent enzyme active site models.
• Benchmarking of various Density Functional Approximations (DFAs) against the generated high-level reference data.
• Re-evaluation of DFA performance on the ENZYMES22 dataset of organic enzyme active site models using CPS-extrapolated reference values.

Main Results:

• While CPS extrapolation altered absolute deviation magnitudes for DFAs, relative rankings remained largely consistent.
• Differences in DFA performance were more pronounced for metalloenzymes compared to organic enzymes.
• The use of CPS extrapolations for reference values had a negligible impact on the overall benchmarking outcomes for organic enzymes.

Conclusions:

• CPS(5,6) extrapolation is recommended as a practical alternative to standard TightPNO settings for generating DLPNO-CCSD(T) reference values.
• The improved accuracy of DLPNO-CCSD(T) with CPS extrapolation enhances its applicability for benchmarking larger organic enzyme models.
• Updated DLPNO-CCSD(T1)/CPS(6,7) energies for the ENZYMES22 set are provided as improved reference data.