Date of Award:

5-2009

Document Type:

Thesis

Degree Name:

Master of Science (MS)

Department:

Computer Science

Advisor/Chair:

Minghui Jiang

Abstract

We present a new method for predicting the secondary structure of RNA sequences. Using our method, each RNA nucleotide of an RNA Sequence is represented as a point on a 3D triangular lattice. Using the Simulated Annealing technique, we manipulate the location of the points on the lattice. We explore various scoring functions for judging the relative quality of the structures created by these manipulations. After near optimal configurations on the lattice have been found, we describe how the lattice locations of the nucleotides can be used to predict a secondary structure for the sequence. This prediction can be further improved by using a greedy, 2-interval post-processing step to find the maximum independent set of the helices predicted by the lattice. The complete method, DeltaIS, is then compared with HotKnot, a popular secondary structure prediction program. We evaluate the relative effectiveness of DeltaIS and HotKnot by predicting 252 sequences from the Pseudobase Database. The predictions of each method are then scored against the true structures. We show DeltaIS to be superior to HotKnot for shorter RNA sequences, and in the number of perfectly predicted structures.

Share

COinS