Characterization of folding pathways within fold families

Proteins fold from a partially structured denatured state to a fully folded native ensemble. Structurally similar proteins need not share significant sequence identity. Consequently, a large number of protein sequences can fold into a relatively small number of protein topologies. We derive a consensus domain dictionary from three publicly available domain dictionaries. The consensus folds (or 'metafolds') from this domain dictionary were scanned for autonomous protein domains. A large fraction of the metafolds from this domain dictionary consisted solely of domains that were not self-contained. We used the selected domains as the basis for high-throughput molecular dynamics simulations of protein native states and unfolding. This set of simulations constitutes the largest set of simulations of protein folding and unfolding to date We validate the native simulations using structural measures and comparison to experiment. We perform an in-depth case study of folding of the three-helix bundle homeodomains, one of the most populated metafolds in our consensus domain dictionary. Finally, we validate the simulations of Pit-1 HD against experimental kinetic studies of folding.