Research On Nucleic Acids Structure Prediction

Nucleic acid is one of the basic substances in the cell,and predicting the tertiary structure and secondary structure of nucleic acid contributes to study of their function.In this paper,we've made an in-depth study on the structural prediction of nucleic acids,including RNA tertiary structure prediction,RNA tertiary structure optimizing,RNA tertiary structure scoring,DNA tertiary structure prediction and RNA secondary structure prediction.We've done a preliminary study on the non-coding RNA tertiary structure prediction,and proposed a fragment assembly based method of RNA tertiary structure prediction.Our first work in this paper is to upgrade the function of 3dRNA,including improving secondary structure tree,extending templates library,designing sampling algorithm,designing automatic loop templates building method and so on.We tested the improved 3dRNA in a test set containing 32 RNA.The results showed that the average RMSD of the predicted results of new 3dRNA was reduced from 3.94 A to 3.24 A,indicating that 3dRNA was further improved.We also validated the ability of 3 dRNA to reconstruct non-Waston Crick base pairs.When the exact templates can not be found,the prediction accuracy of 3dRNA will be greatly reduced.We have therefore developed a program based on the coarse-grained model and using the Monte Carlo simulation to optimize RNA tertiary structure.We designed a new energy function that allows utilizing the results of direct coupling analysis(DCA)as constraints.We compared the optimization program with two other known methods,and the RMSD of our method is 3-6 A lower than theirs.We also devised a knowledge-based all-atom statistical potential(3dRNAscore)for RNA tertiary structure evaluation.It combines distance energy potential and torsion angle based energy potential.In a test set containing 85 RNAs,3dRNAscore could recognize the native structure of 84 RNAs.At present,there are many methods of RNA tertiary structure prediction,but there is no way to predict DNA tertiary structure,so we designed a method for DNA tertiary structure prediction.The method first constructs an initial structure,and then uses the optimization method we designed to optimize the initial structure.Our method not only predicts the tertiary structure of non-helical double-stranded DNA,but also predicts the structure of triple-stranded DNA and G4 DNA.We tested the accuracy of the three types of DNA,respectively.The results showed that the accuracy of predicting the tertiary structure of triple-stranded DNA and G4 DNA was about 4 A,and the accuracy of double stranded DNA was about 8 A.Finally,we designed an iterative dynamic programming algorithm to use the results of direct coupling analysis to predict RNA secondary structure.We compared our methods with other methods.In a test set containing 1242 RNAs,the STY of 796 RNAs was predicted to rank first,with an average of SYT of 0.731;the MCC of 526 RNAs was predicted to rank first,with an average of MCC of 0.683.We have built a corresponding websie for each method,e.g.http://biophy.hust.edu.cn/3dRNA,and all the methods are integrated in package NSP:https://github.com/hust220/nsp.