Canonical analysis of correlated atomic motions in DNA from molecular dynamics simulation

We report a method for analyzing atomic correlated motions in biopolymers from trajectories obtained by molecular dynamics simulation. A correlation coefficient based on the canonical analysis of data is defined which is independent on the relative orientation of atomic displacement. To illustrate the method we studied correlation between positional fluctuations of protons in the double-stranded self complementary oligonucleotide d(CTGAT-CAG), deduced from a 200 ps molecular dynamics simulation in the presence of explicit water molecules and counterions. It is found that on this time scale the motions of protons belonging to different residues are poorly coupled while the motion of a base proton is correlated to the motion of the sugar ring protons of the same nucleotide. Such a method may be generalized to study correlated motions of two distinct domains of a macro-molecule.