Improving Far-Uv Cd Prediction With The Dipole Interaction Model

The dipole interaction model, an implementation of classical electromagnetic theory, is assembled in a computer package: DInaMo/CDCALC. This package is very successful in predicting the far-UV CD π-π∗ transition of proteins, replicating the peak positions, intensity and morphology. The major limitation of the method has been the neglect of the n-π∗ transition. Herein, initial parameters, including the n-π∗ transition, are incorporated with DInaMo/CDCAL. Furthermore, new mean polarizability parameters are developed for the CH3, CH2, and CH groups. Protein structures obtained from the Protein Data Bank (PBD) are energy minimized with NAMD. The energy minimized structures are further modified in six different ways: (1) The methyl (CH3) hydrogens are deleted prior to calculation. (2) Both the methyl (CH3) and methylene (CH2) hydrogens are deleted. (3) All the CH3, CH2, and CH hydrogens are deleted leaving only amide hydrogens. (4) New mean polarizability parameters are implemented only for CH3. (5) New mean polarizability parameters are implemented for both CH3 and CH2 groups. (6) New mean polarizability parameters are implemented for CH3, CH2, and CH groups. There seems to be a correlation between the protein secondary structureal and the kind of modification used for the CD prediction. The β-sheets give best results with including only amide hydrogens while the α-helical proteins give better results ignoring CH3 and/or CH2 groups. Calculations with using new mean polarizability parameters remove the need for different π-π∗ transition parameters and improve the CD results in lower RMSDs and better spectra morphology. In addition, the new parameters for the n-π∗ transition yield normal modes in the correct region and sign for this transition.