A Robust Gene selection Method for Microarray-based Cancer Classification

Abstract

Gene selection is of vital importance in molecular classification of cancer using high-dimensional gene expression data. Because of the distinct characteristics inherent to specific cancerous gene expression profiles, developing flexible and robust feature selection methods is extremely crucial. We investigated the properties of one feature selection approach proposed in our previous work, which was the generalization of the feature selection method based on the depended degree of attribute in rough sets. We compared the feature selection method with the established methods: the depended degree, chi-square, information gain, Relief-F and symmetric uncertainty, and analyzed its properties through a series of classification experiments. The results revealed that our method was superior to the canonical depended degree of attribute based method in robustness and applicability. Moreover, the method was comparable to the other four commonly used methods. More importantly, the method can exhibit the inherent classification difficulty with respect to different gene expression datasets, indicating the inherent biology of specific cancers.

Ensemble of data mining methods for gene ranking

“…The average accuracy of a class recognition in the prostate tumor problem calculated over 100 cross validation runs was equal 92.4%. This result is in a good relation to the most recent results for similar data of the prostate tumor [11], where the declared average accuracy on the PR data gathered in the base [16] was equal 90.98%.…”

“…They rely on different principles and possibly generate different results for the same data sets. A good practice in ill defined problems is application of few methods simultaneously and draw the final conclusion by considering the results of all of them [3,4,11]. Conflicting results in repeated experiments are resolved through attention to the statistical details.…”

“…DNA microarray represents now a powerful tool in biomedical discoveries [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]. They store the expressions of thousands of individual genes on a single surface of the size of the microscope slide.…”

Data mining methods for gene selection on the basis of gene expression arrays

“…This particular problem was considered, for example, by Wang and Gotoh (2010) by using rough set theory, or by Wiliński and Osowski (2012) by using several methods of selection. The issue is a typical feature selection task of data mining (Duda et al, 2003;Guyon and Elisseeff, 2003;Tan et al, 2006).…”

“…Present approaches to this task include various clustering methods (Eisen et al, 1998;Herrero et al, 2001), application of neural networks and support vector machines (Guyon et al, 2002;Huang and Kecman, 2005;Wiliński and Osowski, 2012), statistical tests (Baldi and Long, 2001), linear regression methods applying forward and backward selection (Huang and Pan, 2003), fuzzy logic based algorithms (Woolf and Wang, 2000), rough set theory (Wang and Gotoh, 2009;2010;Świniarski, 2001), various statistical methods (Mitsubayashi et al, 2008;Golub et al, 1999), as well as a fusion of many selection methods (Wiliński and Osowski, 2012;Yang, 2011). Although the progress in this field is fast, there is still a need for better understanding and improvement of the research.…”