Protein-DNA interactions are vital for living organisms. From viruses to humans, the interactions of these two different classes of biopolymers control processes as important and diverse as the expression of genes and the replication, rearrangement, and repair of DNA itself. These processes occur with exquisite specificity. It is still not well understood, however, how this specificity occurs. How does a given protein biopolymer choose one particular sequence of DNA bases over another?

The past decade has seen a great increase in the number of known protein sequences and determined structures of protein-DNA complexes. To further understand how proteins specifically target DNA, we use both of these data sources. First, using protein amino acid sequences, we study the level of sequence similarity needed to confidently predict whether two sequences have the same or different specificities for DNA. We find that the level depends on the group of proteins being studied, but that a technique motivated by percolation theory can determine the similarity level that is needed. This work also leads to the prediction of proteins that carry out unusual functions, such as very different specificities or modes of binding.

Second, we use the groupings of DNA-binding proteins to predict which amino acids have been used to make a protein specific for one DNA sequence while a related protein is specific for another DNA sequence. These predictions can guide experimental studies as well as provide starting points for design.

Finally, we consider how crystal structures can be used to describe the energetics of other protein-DNA complexes. Using methods originally developed to describe protein folding, we find the method that best uses existing structural information to determine the energetics of DNA binding.

The study of protein-DNA specificity will continue for many years, but there are now more tools available to describe these vital interactions.