Automated RNA Tertiary Structure Prediction
Matthew G. Seetin and David H. Mathews
Department of Biochemistry and Biophysics and Center for RNA Biology, University of Rochester Medical Center
A novel protocol employing simulated annealing and steered molecular dynamics for all-atom RNA tertiary structure prediction is presented. The restraints are derived from secondary structure, coaxial stacking predictions for helices in junctions, co-variation analysis, and, when available, cross-linking data. The protocol is tested on the Alu domain of the mammalian signal recognition particle RNA, the tRNAPhe of Saccharomyces cerevisiae, the hammerhead ribozyme, and the hepatitis C virus internal ribosomal entry site. The resulting pool of decoy structures is scored with a scoring function that maximizes the radius of gyration and number of base-base contacts to identify the best predicted structure. These predictions are compared with the crystal structures using both root mean square deviation (RMSD) and the accuracy of base-base contacts. This simple ab initio approach using the above restraints is sufficient to make accurate predictions of the structures, with, for example, an RMSD of 5.0 Angstroms for all heavy atoms in the mammalian SRP Alu domain.