New version of the theoretical databank of transferable aspherical pseudoatoms, UBDB2011--towards nucleic acid modelling.

The theoretical databank of aspherical pseudoatoms (UBDB) was recently extended with over 100 new atom types present in RNA, DNA and in some other molecules of great importance in biology and pharmacy. The atom-type definitions were modified and new atom keys added to provide a more precise description of the atomic charge-density distribution. X-H bond lengths were updated according to recent neutron diffraction studies and implemented in the LSDB program as well as used for modelling the appropriate atom types. The UBDB2011 databank was extensively tested. Electrostatic interaction energies calculated on the basis of the databank of aspherical atom models were compared with the corresponding results obtained directly from wavefunctions at the same level of theory (SPDFG/B3LYP/6-31G** and SPDFG/B3LYP/aug-cc-pVDZ). Various small complexes were analysed to cover most of the different interaction types, i.e. adenine-thymine and guanine-cytosine with hydrogen bonding, guanine-adenine with stacking contacts, and a group of neutral and charged species of nucleic acid bases interacting with amino acid side chains. The energy trends are well preserved (R(2) > 0.9); however the energy values differ between the two methods by about 4 kcal mol(-1) (1 kcal mol(-1) = 4.184 kJ mol(-1)) on average. What is noticeable is that the replacement of one basis set by another in a purely quantum chemical approach leads to the same electrostatic energy difference, i.e. of about 4 kcal mol(-1) in magnitude. The present work opens up the possibility of applying the UBDB2011 for macromolecules that contain DNA/RNA fragments. This study shows that on the basis of the UBDB2011 databank electrostatic interaction energies can be estimated and structure refinements carried out. However, some method limitations are apparent.