Advances in sequencing technology lead to rapidly growing amount of RNA sequence information and in the mean- time, create an increasing gap between the number of known sequences and the number of known structures. Moreover, the dramatic increase in the amount of non-coding RNAs and the discovery of their functions require more than ever a clear understanding of RNA structures. However, experimental determination of RNA structure is time consuming and cannot keep up the pace with ever-increasing demand. This causes a pressing demand to develop accurate computational models to predict RNA 3D structures. In the past ten years, remarkable progress has been achieved on RNA structure predictions. Further advances of RNA structure prediction, however, are blocked by two main hurdles: (a) the inability to predict long-range tertiary contacts and (b) lack of structural templates. Building upon our previous highly successful coarse-grained RNA folding model (Vfold model), we propose a new approach to tackle these challenges and to develop a new RNA structure prediction model. We have three major aims in this proposal: (a) To systematically develop a method to calculate tertiary folding entropy and free energy. (b) To develop a free energy-based approach to predict the base pairs and the tertiary contacts from the sequence. (c) To develop a three-dimensional all-atom model and to systematically test and refine the model based on the experimentally determined structures. We will also continue the improvement and dissemination of our Vfold web server to best serve the scientific community. Our proposed method will integrate a physics-based modeling of entropy and free energy of tertiary folds with knowledge-based training of scoring functions. Key advantages of the approach are that it is based on the complete conformation ensemble instead of randomly sampled conformations and that the scoring function accounts for not only the native folds but also the effect from the nonnative interactions. Other advantages include the use of an electrostatic model that can treat Mg2+ ions and the inclusion of hydration energy in the selection of structural models. The preliminary tests show very promising results, suggesting the feasibility of our approach. Furthermore, through collaborations with biochemists and RNA-based cancer biologists, we will continuously test, refine and validate the model, and apply the model to solve biologically significant and timely problems such as RNA-based therapeutic design.
This project will develop a model for accurate predictions of all-atom structures for RNA tertiary folds. This predictive model will contribute to the quantitatie understanding of RNA cellular functions and to the rational design of RNA-based therapeutics.
Showing the most recent 10 out of 74 publications
