PREDICTING PROTEIN-DNA INTERACTIONS USING GEOMETRIC DOCKING - CAN DNA SPECIFICITY IMPROVE RESULTS?

Protein-DNA interactions are involved in many essential biological activities. Since there is no simple mapping code between DNA base pairs and protein amino acids, the prediction of Protein-DNA interactions is a challenging problem. Existing geometric docking algorithms between protein and DNA are based on the assumption that there is a shape complementarity between the participating molecules and consider slightly the specificity between the protein and DNA sequences. In order to test whether a systematic consideration of DNA specificity might improve the docking algorithms result, sequence - dependent DNA shape models were used for docking. A non-redundant dataset of 169 protein-DNA complexes was selected from PDB. Three DNA shape models were docked, by PatchDock algorithm¹̛ , to each of the 169 proteins in the dataset using three DNA shape models. The first DNA model was the DNA structure taken from PDB, the second DNA shape model was built using a coarse-grained molecular dynamic simulations model (to study sequence-dependent DNA shape) and the third DNA shape model was built as a standard B-DNA using X3DNA software. All three DNA structural models were based on the same DNA sequence identical to the DNA in the PDB complex. In 150 out of the 169 (89%) PDB complexes in dataset, docking a sequence dependent DNA model produced better results than the regular B-DNA. The results of docking a sequence-dependent DNA shape make it clear that the specificity of protein-DNA interactions could contribute to finding the protein-DNA binding site. The insights from such a study could help to characterize the unique properties of protein-DNA interfaces and identify new drug target sites.

References

1. Duhovny D., Nussinov R. and Wolfson HJ. Nucleic Acids Res 33, W363-7 (2005).