Some Evaluations about Gene Expression Data Clustering

• Main Authors: Jeng Yuan Cheng, 鄭景元
• Other Authors: Chuan Yi Tang
• Format: Others
• Language: en_US
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/55435801211871468902

碩士 === 國立清華大學 === 資訊工程學系 === 91 === The microarray experiments result in a great quantity of data. The researchers have attempted to find the information of the data by different clustering methods. Methods to evaluate the gene-expression data clustering are targeted. The expressive patterns of several genes frequently cause the contradiction of distance relation in the data set. We aim at a method to filter the genes that often generate the contradiction before clustering. The similarity of two genes must be known before clustering. The distance function represents the similarity of two genes. A method is proposed to evaluate the distance functions which evaluate the similarity of two genes to help the researchers select an appropriate distance function for the data set. The aforementioned two methods described above are based on the identifiable data from the published literature. The distance function and clustering algorithm produce the clustering result. This paper presents a method to evaluate the clustering algorithm on the basis of the same distance function. This method is also capable of listing the genes with low confidence. No matter which cluster the genes are classified into, the researchers can have a reference about the clustering result.