|
Protein, DNA and RNA are important biomolecules. More than100years research found that RNA played an essential role in many biological processes, like protein synthesis, biological catalysis and genetics. Since the structure determines the function, studies on primary sequence, secondary structure and tertiary structure of RNA draw more attention. Because of the difficulty of crystallization, limitation of experimental technology and high cost of measurement, predicting RNA structures theoretically is becoming more and more popular. So far, most of the tertiary structure predictions are based on the secondary structure of RNA. And more or less, the function of RNAs can be known from the secondary structure. Therefore, studying on RNA secondary structure is quite important and meaningful.There are many methods for RNA secondary structure prediction, including comparative sequence analysis, dynamic programming based on minimal free energy, combinatorial optimization and heuristic algorithms. Among these methods, comparative sequence analysis is the best in prediction accuracy. However, comparative sequence analysis requires high identity with homology sequences and sequence alignment affects the prediction accuracy directly. The small number of known RNA structures prevents this method from being widely used. Except comparative sequence analysis, dynamic programming based on minimal free energy prevails in accuracy and is the most popular method. However, its accuracy decreases dramatically as the length of RNA increases and it has weak power to predict pseudoknots. Combinatorial optimization and heuristic algorithms are novel methods, but their accuracy and convergence are the obstacles that prevent them from being widely used. Here, we introduce a new method based on minimal free energy and conservation of hairpin structures. This method can predict RNAs with relatively long sequence and it’s accuracy is better for some of the RNAs.
|