Microarray Probe Design Using ǫ-Multi-Objective Evolutionary Algorithms with Thermodynamic Criteria Soo-Yong Shin, In-Hee Lee, and Byoung-Tak Zhang Biointelligence Laboratory, School of Computer Science and Engineering, Seoul National University, Seoul 151-742, Korea {syshin, ihlee, btzhang}@bi.snu.ac.kr Abstract. As DNA microarrays have been widely used for gene expres- sion profiling and other fields, the importance of reliable probe design for microarray has been highlighted. First, the probe design for DNA mi- croarray was formulated as a constrained multi-objective optimization task by investigating the characteristics of probe design. Then the probe set for human paillomavrius (HPV) was found using ǫ-multi-objective evolutionary algorithm with thermodynamic fitness calculation. The evo- lutionary optimization of probe set showed better results than the com- mercial microarray probe set made by Biomedlab Co. Korea. 1 Introduction DNA microarray, especially oligonucleotide array, consists of the DNA sequences called probes, which are DNA complementaries to the genes of interest, on a solid surface. When the molecules of a cell is put to the microarray, if there exists a complementary oligonucleotide to one of the probes, it would hybridize to the probe so that a user can detect it using various methods. In this way, DNA microarray can provide the information on whether a gene is expressed or not for hundreds of genes simultaneously. Therefore, DNA microarray is widely used to study cell cycle, gene expression profiling, and other DNA-related phenomena in a cell; and has become the method of choice to monitor the expression level of a large number of genes. By the way, microarray depends on the quality of probe sets that used. If a probe hybridizes to not only its target gene but also other genes, the microarray may produce misleading data. Thus, one needs to design the probe set care- fully to get precise data. Till now, lots of probe design methods and strategies are suggested reflecting its importance [16]. Gordon and Sensen proposed a Os- prey system based on various well-defined criteria [5]. Zuker group implemented OlgioArray 2.0 using thermodynamic data to predict secondary structures and to calculate the specificity of targets on chips [10]. Wang and Seed suggested OligoPicker which uses BLAST search for sequence specificity decision [18]. F. Rothlauf et al. (Eds.): EvoWorkshops 2006, LNCS 3907, pp. 184–195, 2006. c  Springer-Verlag Berlin Heidelberg 2006