A method to improve structural modeling based on conserved domain clusters
Fa Zhang, Lin Xu, Bo Yuan
Abstract
Fa Zhang, Lin Xu, Bo Yuan
Abstract
Homology modeling requires an accurate alignment between a query sequence and its homologs with known three-dimensional (3D) information. Current structural modeling techniques largely use entire protein chains as templates, which are selected based only on their sequence alignments with the queries. Protein can be largely described as combinations of conserved domains, and already more than two-third of the known protein domains can be found in the protein data bank (PDB). We presented a method to improve structural modeling based on conserved domain clusters. First, we searched and mapped all the inter pro domains in the entire PDB, partitioned and clustered homologous domains into the domain-based template library. For each of the resulting clusters created, a multiple structural alignment was generated based only on the 3D coordinates of all the residues involved. Then we used the structural alignments as anchors to increase the alignment accuracy between a query and its templates, and consequently improve the quality of predicted structure for query protein. We implemented the method on DAWNING 4000A cluster system. The preliminary results show that our domain-based template library and the structure-anchored alignment protocol can be used for the partial prediction for a majority of known protein sequences with better qualities.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Homology modeling requires an accurate alignment between a query sequence and its homologs with known three-dimensional (3D) information. Current structural modeling techniques largely use entire protein chains as templates, which are selected based only on their sequence alignments with the queries. Protein can be largely described as combinations of conserved domains, and already more than two-third of the known protein domains can be found in the protein data bank (PDB). We presented a method to improve structural modeling based on conserved domain clusters. First, we searched and mapped all the inter pro domains in the entire PDB, partitioned and clustered homologous domains into the domain-based template library. For each of the resulting clusters created, a multiple structural alignment was generated based only on the 3D coordinates of all the residues involved. Then we used the structural alignments as anchors to increase the alignment accuracy between a query and its templates, and consequently improve the quality of predicted structure for query protein. We implemented the method on DAWNING 4000A cluster system. The preliminary results show that our domain-based template library and the structure-anchored alignment protocol can be used for the partial prediction for a majority of known protein sequences with better qualities.
Key concepts: Protein Data Bank (RCSB PDB), Structural alignment, Template, Computer science, Protein Data Bank, Sequence alignment, Multiple sequence alignment, Homology modeling