Software for the accurate de novo 3D structure prediction of RNA

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The discovery of new and interesting RNA sequences from genome sequencing projects, and the urgency to unravel their functions, has led to a dramatic push for novel structural determination techniques. Current experimental methods for three dimensional structure determination of nucleic acids such as x-ray crystallography and NMR cannot keep pace with the day to day discovery of sequences that need representative structures to be solved or modeled. Thus, there is a clear need to develop tools for 3D structure prediction given only the primary sequence and when available, experimental constraint information. Compared to proteins, RNA structure prediction has received limited resources, and only recently has the field gained attention by the scientific community. As such, RNA prediction has largely relied on protein prediction methodologies despite the vast intrinsic differences between proteins and nucleic acids. Although many of these tools have shown significant advances in the prediction quality, they have also demonstrated low reliability and are often limited to prediction of very small RNAs. In addition, the majority are either manual or semi-automated, which requires an experienced user and a variety of intermediate software packages. To address such concerns, DNA Software, Inc. (DNAS) has developed an RNA homology modeling software, NA-CAD (Nucleic Acid Computer Aided Design) that has a unique force field specifically optimized for RNA. This tool has demonstrated success in homology modeling of large RNA-protein complexes such as the small ribosomal subunit of Pseudomonas aeruginosa. We would now like to extend NA-CAD to include a component for de novo structure prediction. The force field in NA-CAD and the free energy based secondary structure prediction algorithm in our flagship software product Visual OMP provide an advantageous starting point for developing a unified tool that can accurately predict de novo the tertiary structure of medium to large RNA targets. This proposal addresses the engineering of novel algorithms for handling difficult structural motifs such as multiloops, pseudoknots, and multiple domains and incorporating experimental constraints to improve prediction quality. Additionally, coarse-grained models for representing RNA residues and accelerated classical molecular dynamics simulations will be implemented to increase conformational sampling in a tractable computational time frame. PUBLIC HEALTH RELEVANCE: We propose to develop an accurate, fast, and unified de novo structure prediction tool optimized for medium to large sized RNAs. This tool will generate valuable structural models that will help elucidate the functions of RNAs that do not have solved crystal or NMR structures. The proposed de novo tool will be incorporated into our homology modeling software, NA-CAD, so that it will be able to generate three-dimensional homology models of pharmaceutically relevant RNA targets and to model potential drug-resistant mutants, which will be beneficial to researchers involved in structure-based drug discovery.