Optimizing Genetic Algorithm For Motif Discovery

Understanding the regulation of the gene expression is one of the challenge problems in bioinformatics and molecular biology. Transcription regulation is a key step in the regulation of gene expression, transcription factor binds the transcription factor binding sites in the gene promoter sequences, starts the gene transcription and controls the efficiency of gene transcription. As a result, predicting transcription factor binding sites is an important part of the study of gene regulation transcription, which accelerates the research on the transcriptional regulation of gene expression and the construction of transcriptional regulatory network.In this thesis, The Genetic Algorithm via OPTimization(GAOPT) is proposed. First we focus on discussing the motif model, and use the vector to describe motifs. Then we analyse and explain the Bayesian scoring function for motifs, which is used as the fitness function in GAOPT. At last, In order to make genemic algorithm have a good start, GAOPT provide a solution space of possible motifs generated by random projection instead of the one generated by random.Moreover, GAOPT also incorporates two additional operators to optimize the optimal motif. Experimental results demonstrate that GAOPT performs well on both simulated and biological samples.