Research On Hte Microarray Based Gene Mining Algorithm

As the rapid development of the gene chip technology, the biology research method has been greatly changed. The experiment based qualitative research is moving towards the data based quantitative research slowly, which give birth to the bioinformatics and make the traditional biology more and more powerful. Especially the recent research focusing on the microarray have the great power on the ability of solving the biological problem.The research direction of this dissertation is about the microarray based gene mining in the bioinformatics subject. The content of this research and result is as follow: firstly, the theory on the relevance and the feature gene mining is summarized, which form the basis of this reseach. Secondly, a measure of a gene set is proposed for evaluating the expression level difference of the genes in the set, which is a combination of the fisher linear classification and the t-statistic method and give a relevance value to every gene in the set as well as to the whole gene set. Thirdly, in the traditional decision tree based ensemble gene mining method, the gene number in the node of the tree is constrained to one. A modified algorithm is proposed, which uses the character of the FLDT measure to realize the extension that the number of the gene contained in every node can be any number smaller than N (N is specified by user). The extension cancels the traditional algorithm's constraint and makes the algorithm more flexible and more powerful on the classification. Fourthly, there are many gene mining algorithm presently and the method evaluation is becoming a research direction. The evaluation methods proposed mainly based on the mathematical analysis or the random simulation. In this dissertation, a pseudo-relevance gene set based method is proposed to evaluate the gene mining method or its mined gene list and accordingly we propose a multiple gene list fusion algorithm IndexFusion, which try to get a better gene list from several worse ones. Finally,a gene mining system is constructed, whitch contains all the algorithm proposed in this desseration and other methods.