Next: 2. Materials -- Up: Prediction of RNA Previous: Prediction of RNA

1. Introduction

econdary structure prediction for RNA is fundamentally different from three dimensional molecular modeling [1]. A secondary structure of an RNA molecule is simply a collection of predicted base pairs subject to a few simple rules. Base pairs can be either G-C or A-U Watson-Crick pairs, or the weaker G-U pair.

Secondary structures can be used in part to explain translational controls in mRNA [2], and replication controls in single-stranded RNA viruses [3]. Secondary structure modeling can also be used as a first step to the more intricate process of three dimensional modeling [4]. This could include the modeling of ribosomal RNA or catalytic RNA's such as group I introns.

When numerous homologous RNA's are available, secondary structure prediction is possible using the comparative sequence analysis method [5] Though largely a labor intensive method when first used, it is now being made more automatic [6]. The comparative approach requires the RNA sequences to be similar enough so that they can be reliably aligned.

When one wishes to fold a single sequence, the usual method is to predict a secondary structure that is optimal in some sense. Recursive algorithms borrowed from the area of sequence alignment have been modified and applied to the RNA folding problem by a number of people [7,8,9,10,11,12,13]. While some early methods ( e.g.[9]) maximized the number of base pairs in a secondary structure, the later methods have all attempted to solve the problem by minimizing a computed free energy.

The method described here [14,15,16] uses energy rules developed by Turner and colleagues [17,18,19]. Negative stabilizing energies are assigned to the stacking of base pairs in helical regions, and to single bases that stack at the ends of helical regions. Otherwise, destabilizing energies are assigned to bulge, interior, hairpin and multi-branched loops. The energies of base stacking and the destabilizing loops are assumed to be additive in computing the overall energy. Pseudoknots, described in [20], are not permitted in secondary structure calculations.

Michael Zuker
Thu Nov 2 14:28:14 CST 1995