| With the carrying out and implementation of the Human Genome Project(HGP),huge biomolecular sequence, structure and function data appears and growsgeometrically. And with the rapid development of computer technology, people havebegun to better analyze and handle these substantial growth data, Thus, bioinformaticsemerged. By the rapid development of bioinformatics and people gradually in-depthstudy of RNA, now it is not only DNA’s age, but also for RNA’s important role in thelife, RNA has taken a lot of people’s attention. RNA molecule in a cell is now playingas a carrier of genetic information, catalyst in cell processing and the media ofdecision the level of gene expression and so on, and gradually takes the similar statusin the center law with DNA and protein. The function and structure of RNA is closelyrelated, research on the relationship between the function and structure of RNA is ofgreat importance to reveal the role of RNA molecules in the life, but also it isimpportant for the development of the entire genome project. Conventional X-ray andNMR biological test method to study RNA structure is costly and time-consuming, bythe computer method, using theoretical guidance to predict RNA structures, hasbecome an important research in molecular biology. also, people continue to improvethe old prediction algorithm and re-establish a complete new RNA structureprediction model, it has been becoming a hot topic in bioinformatics.Now pseudoknot structure prediction algorithms mainly include three categories:one is comparative sequence analysis method based on homologous sequences, one isdynamic programming algorithm combined free energy model, one is heuristicalgorithm by using locally optimal solution to determine the global optimal solution.These algorithms combine various disciplines methods to predict RNA secondarystructure, and have achieved very good results, but no matter what method, they allhave certain shortcomings, especially for the pseudoknot prediction, it did not reachsatisfactory results, therefore, constructing a rational and efficient prediction algorithm containing pseudoknot remains an open and an important problem.Every method now could not accurately predict pseudoknot, mainly attributed tothe following two reasons: First, these methods ignore the folding mechanism ofpseudoknot, second, the thermodynamic models is not so accurated. Studies haveshown that the folding of pseudoknot may not simultaneously, that is, first it formspseudoknot-free structure, and then further interacts on the basis of the secondarystructure’s local motif, finally formed RNA secondary structure with pseudoknot.Based on the above theory, this paper presents a new method LIFold to predict RNAsecondary structure with pseudoknot, for a given RNA sequence, the algorithm firstapplied the dynamic programming algotithm to get the structure without pseudoknot,this process is based on Mathews-Turner energy model parameters, the predictedstructure conformed to the principle of energy minimization. Then is the pseudoknotprediction, this process mainly covers building a new pseudoknot energy model andadjusting the pseudoknot stem. Finally, using the energy model to screen pseudoknotstem, if the free energy of the overall structure without pseudoknot decreased, thentemporarily stored the pseudoknot stem, until all of the stems were calculated, thenselected the optimal energy structure with pseudoknot as the final result.The method LIFold was first based on HotKnots data set to test, and comparedwith HotKnot, ILM, PknotsRG and IPknot algorithms, LIFold obtained the optimalprediction result, sensitivity and specificity achieved84%and80%respectively. Thenevaluated the method in the PK168set, after compared with HotKnot, pknotsRG, ILMand FlexStem algorithms, the sensitivity of LIFold is78%,only less than the optimalsensitivity two percentages, specificity gained the optimal result73%. Therefore, thismethod not only achieved good results in accuracy but also its stability was in a verygood performance. |
PDF Full Text Request