A crystallographic approach to DNA bending: prediction of nucleosome formation by DNA triple repeats and other repetitive sequences.

DNA bending is due to two main factors: the inherent curvature of the sequence and its flexibility. Most methods of analysis do not allow a differentiation between these two factors. In this paper I show that the flexibility of DNA sequences can be estimated from the standard deviation of roll values determined by X-ray crystallography for each base step. As an application of this approach, the nucleosome formation ability of triple repeat sequences has been determined and shown to be in agreement with the experimental results. Local variations in twist do not appear to have any influence on nucleosome formation.