The Application Of Partial-ordered Theory In Knowledge Discovery Of Gene Experssion Data

With the rapid development of biotechnology and computer technology, mankind hasentered a highly information-oriented era. During this process, numerous biological datawas produced. To discover valuable rules by mining and utilizing the massive biologicalinformation has become an important research field of biology and medicine.Formal Concept Analysis theory is based on mathematical concepts and conceptuallevel, can be used as a algorithm of data mining. On this basis, professor Hong Wenxuepresents the construction theory of the structural partial-ordered attribute diagram, its mainadvantage is that the inherent logic and organization structure of data can be fullyillustrated, so as to provide a system visualization tool for analyzing the associationsbetween data.Based on Formal Concept Analysis theory, the hierarchical optimization method offormal context and the construction theory of partial-ordered attribute diagram are raised.In this paper, a new method for visually discovering knowledge of gene expression data ispresented.In this paper, theory of the partial-ordered attribute is applied to gene expression datamining. The data used here include human multiple tissues microarray data and breastcancer microarray data. The data is standardized and discretized, and specific genesassociated with breast cancer were flitered out. Respectively using the binary relations ofgenes and tissues, genes and the breast tissue samples as the application backgrounds togenerate the formal context and build the structural partial-ordered attribute diagram. Byanalyzing the attribute diagram, specific expression patterns of breast cancer specificgenes and genes differentially expressed in tumor tissue were found. It provides aneffective approach for the discovery of unknown specific genes associated with kinds ofdiseases.