Prediction of molecular structure and function from sequence information has become increasingly possible with the rapidly growing abundance of sequence data. The most successful prediction methods have been comparative methods, which use alignments of homologous sequences to detect positions in the sequence that evolve in tandem, and are therefore likely to be interacting. I intend to improve the accuracy of these methods by incorporating the phylogenetic relationships among the sequences into these analyses. I will then use these more accurate structural predictions to make better multiple sequence alignments and phylogeny estimates. Finally, I plan to tie structure prediction, sequence alignment, and phylogeny estimation together in an iterative fashion in order to improve all three procedures.
