Particle swarm optimization for solving DNA sequence design problem

• Main Author: khalid, Noor khafifah
• Format: Thesis
• Language: English
• Published: 2010
• Subjects: Q Science (General)
• Online Access: http://eprints.utm.my/id/eprint/12748/1/NoorKhafifahKhalidMFKE2010.pdf; http://eprints.utm.my/id/eprint/12748/
• Institution: Universiti Teknologi Malaysia

Deoxyribonucleic Acid (DNA) has certain unique properties such as selfassembly and self-complementary in hybridization, which are important in many DNA-based technologies. DNA computing, for example, uses these properties to realize a computation, in vitro, which consists of several chemical reactions. Other DNA-based technologies such as DNA-based nanotechnology and polymerase chain reaction (PCR) also depend on hybridization to assemble nanostructure and to amplify DNA template, respectively. Hybridization of DNA can be controlled by designing DNA sequences properly. In this thesis, sequences are designed such that each sequence uniquely hybridizes to its complementary sequence, but not to any other sequences. This objective can be formulated using four objective functions, namely, similarity, Hmeasure, continuity, and hairpin. To achieve this, particle swarm optimization (PSO) for DNA sequence design is proposed to minimize the objective functions subjected to two constraints: melting temperature and GCcontent. Two models are developed, namely the Continuous PSO and Binary PSO. Since DNA sequence design is a multi-objective optimization (MOO) problem, two methods to solve MOO are used in this thesis. These methods are the aggregation-based method and criterion-based method, particularly vector evaluated PSO (VEPSO). The implementation of PSO algorithm for DNA sequence design is first started with application of both proposed models to aggregation-based method. Then, the results between these models are compared. It is found that better results are produced by Binary PSO. Next, VEPSO is used to design DNA sequences based on Binary PSO. The results show that several set of good sequences are produced, which are better than other research works where only a set of DNA sequences is generated.