Improved structural modeling based on conserved domain clusters and structure-anchored alignments

In this paper, we presented a method to improve structural modeling based on conserved domain clusters and structure-anchored alignments. We first constructed a template library of structural clusters for all conserved sequence domains. Then, for each cluster, we built the profile using the structure and sequence information. Finally we use the profile and structural alignments as anchors to increase the alignment accuracy between a query and its templates. Our preliminary results show that this method can be used for the partial prediction for a majority of known protein sequences with better qualities.