Perturbation Methods for Obtaining Interactions Energies. Can SAPT2+/jun-cc-pVDZ predict adequate interaction energies when biochemically relevant motifs are present?

In the field of non-covalent interactions, there has always been a great interest in finding the appropriate methodology to analyze bond energies and properties. There are multiple approaches; however, those based on symmetry adapted perturbation theory (SAPT) are interesting for two different reasons: quality of the interaction energy and how it is obtained. Total interaction energies are computed in SAPT as the sum of the electrostatic, repulsive, inductive and dispersive components. This provides enormous information about the intimate nature of intermolecular interactions.

The performance of a variety of symmetry adapted perturbation theory (SAPT) methods for describing non-covalent interactions has been tested in several studies. The appropriate level depends to a certain degree on the nature of the interaction and the nature of the database, however, there is a methodological combination that can be considered as a reference. The SAPT2+(3)δMP2 truncation combined with the aug-cc-pVTZ basis set offers an outstanding performance for the majority of non-covalent complexes. This methodology produces interaction energies of excellent quality with low relative errors and little error spread so it can be adopted as a methodology to obtain reference energies for most applications of interest in chemistry and biochemistry. The problem that SAPT2+(3)δMP2/aug-cc-pVTZ faces is the computational resources demanded. These requirements grow enormously with size so that it soon becomes unfeasible for most systems of interest in biochemistry.

When the computational cost is prohibitively high it has been suggested the use of SAPT2+ level in combination with the jun-cc-pVDZ basis set. This methodology is known to give remarkable results at a reduced computational cost.

In this work, the goodness of the SAPT2+ methodology to produce interaction energies of non-covalent systems is explored using the so-called blind database. This database consists on a set of dimers bearing different type of interactions at equilibrium and non-equilibrium distances. Likewise, the SAPT2+/jun-cc-pVDZ methodology is employed to describe a set of prototypical interactions found in biochemical systems involving sugars, proteins and nucleic acids.