Welcome to RDfolder server for prediction of RNA secondary structure

Introduction

RNAs play an important role in many biological processes. Knowing their structures is important in understanding their functions. In view of the difficulties in the experimental determination of RNA structures, the methods of theoretical prediction for known sequences are often used. In this server, we provide two methods for prediction of RNA secondary structure for molecular biologists, which are random stacking of helical regions (here denoted as the RS method for short) and helical regions distribution (here denoted as the HD method for short). Both the RS and the HD methods are used for prediction of single sequences. A short description on these two methods is given below. See reference (1) for detailed information.

The RS method

The main idea of the RS method is to assume that RNA folding is in the direction of lowering the free energy of the forming structures, but the final structures may not have the minimum free energy. The predicted structures of the RS method are calculated using Monte Carlo simulations. By repeating the simulations, we can obtain many secondary structures for a given sequence. Among these structures, some maybe appear only several times, some maybe appear many times. Here we take the structure with the highest frequency as the dominated structure. Another important parameter in the RS method is the number of simulations, i.e. how many secondary structures should be calculated. According to our previous study, 100 simulations are enough for short sequences such as tRNA molecules. With the increase of sequence length, the highest frequency of the predicted structures will become very low. At this time, the HD method can be used instead.

The HD method

There are three steps to predict secondary structure using the HD method. The first step is to generate a set of secondary structures by the RS method, to find the helical region with the highest frequency, and to add it to the current structure. The second step is to delete all incompatible helices from the helical regions list of the current structure. The third step is to repeat step 1 and step 2 until no helical regions in the list. We take the final current structure as the predicted structure.

Reference

1. Li, W.J. and Wu, J.J. (1998) Prediction of RNA secondary structure based on helical regions distribution. Bioinformatics, 14, 700-706