Sequence-specific thermal fluctuations identify start sites for DNA transcription

We report successful comparisons between model predictions for intrinsic thermal openings and experimental transcription data, showing that large and slow thermally induced openings (bubbles) of double-stranded DNA coincide with the location of start sites for transcription. Investigating viral and bacteriophage DNA gene promoter segments, we find that the largest opening occurs at the transcription start site in all cases studied. Other probable large openings predicted in our model appear to be related to other regulatory sites. The coherent dynamics is determined by a combination of sequence specificity (disorder), nonlinearity, and entropy, controlled by the long-range consequences of local base-pair stacking constraints.