Research Of Mining Frequent Closed Patterns In Microarray Datasets

High-throughout microarray technology, which is different from traditional gene expression detect technology, can detect tens of thousands of genes simultaneously. It provides powerful technical support for the research of gene regulatory network in functional genomics. The association analysis method can be applied to analyzing the microarray datasets, mining association rules of genes, and then constructing gene regulatory network.The mining of frequent closed patterns plays an essential role in association analysis method. Two new improved algorithms were proposed by detecting the disadvantages of recent algorithms applying to microarray datasets for mining frequent closed patterns. The main work can be summarized as follows:1) The principles, processes and basic problems of proposed frequent (closed) pattern algorithms were anatomized. Furthermore, the author analyzed the merits and weakness of these algorithms.2) Based on the row enumeration space, a tree structure, LG-tree was suggested and a new algorithm, MFCPLG was proposed for mining frequent closed patterns. MFCPLG searched the row enumeration space in depth-first, combined the single path pruning, and obtained good performance.3) Based on the row enumeration space, a hyperlink structure, HT-struct was suggested and a new algorithm, HTCLOSE was proposed for mining frequent closed patterns. HTCLOSE adopted depth-first space search strategy, combined efficient pruning and ingenious hyperlink organizing, and obtained even good performance.