Bioinformatic analysis of the protein/DNA interface

Despite extensive studies of the geometry of protein/DNA interfaces the understanding of the affinity and specificity of the protein/DNA interactions remains elusive. We present a novel approach to geometric analysis of protein/nucleic acid interfaces that is based on classification of their local conformations. Protein structures are divided into a series of pentapeptide fragments and each is assigned one of 16 conformers or ‘protein blocks’ [de Brevern et al. Proteins 41, 271 (2000)]. Similarly, each DNA step (unit [base]sugar-phosphate-sugar[base]) is assigned one of ∼20 DNA conformers [Svozil et al. NAR 36, 3690 (2008)]. Significantly, local structures classified into distinct conformers can be represented by symbols so that they can be analyzed more easily than complicated 3D objects. Size of the fragments used for the classification allows analysis of structural features of the interface at a scale intermediate between too detailed interatomic contacts on one side and too crude protein motifs (helix-turn-helix, Zn-finger, ...) on the other. We will present correlations between protein and DNA conformers at their interface from more than 15 hundred protein/DNA crystal structures broken by various criteria of the analyzed structures as protein functional classification, e.g. in GO or Pfam or by crystallographic properties, resolution or overall structure quality.Acknowledgments. This work was supported by Czech-France collaboration Barrande (MEB021032) and Czech Science Foundation (P305/10/2184).