Various computational approaches have been used to model protein-DNA interaction interfaces at an atomic level. We describe their most promising scope of applications and discuss their assets and drawbacks. Structure-based computational methods require three-dimensional protein-DNA complexes gained either by X-ray crystallography or by in silico modeling. Docking approaches, molecular dynamics, and Monte Carlo simulations are promising techniques to model transcription factor-DNA complexes in silico. Both experimentally determined and ab initio designed protein-DNA complexes can be analyzed by statistical methods. We describe the differences of several statistical potentials and how they were obtained. Position weight matrices obtained from structure-based approaches can then be used to scan efficiently and more accurately genome-wide for transcription factor binding sites.
In a case study on WRKY-DNA complexes we present a computational modeling technique for the ab initio design of a specific transcription factor-DNA complex. This procedure is generally applicable to similar problems. The resulting three-dimensional interaction interface provides the basis for studying specific side chain and base interactions. Moreover, the results give hints towards varying specificity and function of different representatives of the WRKY protein family. This study provides valuable insights into the interplay between transcription factors and DNA in three dimensions and opens up new perspectives for their design.