The False Discovery Rate in Simultaneous Fisher and Adjusted Permutation Hypothesis Testing on Microarray Data

Abstract

Background and Objectives: In recent years, new technologies have led to produce a large amount of data and in the field of biology, microarray technology has also dramatically developed. Meanwhile, the Fisher test is used to compare the control group with two or more experimental groups and also to detect the differentially expressed genes. In this study, the false discovery rate was investigated in the simultaneous Fisher and adjusted permutation hypothesis testing on microarray data. Methods: In this study, first, false discovery rate was computed through the simulation study and selection of three different modes for the samples of control group and experimental groups, then, these two methods, were applied to 8799 genes related to brain cell of the 29 rats (in three age groups of young, middle-aged, and aged), and the effect of the process of brain aging, was investigated on increased development of the Alzheimer disease. Results: The results showed that the Fisher permutation methods cannot control the false discovery rate, but the adjusted permutation method works better and detects the significant differences more accurately, therefore, the number of false positives decrease in the second method. Conclusion: Considering the results of this study, use of the customary methods such as the Fisher permutation test, which is the base of analyzing biological data in many of the software, do not have suitable efficiency in the large-scale data, including microarray data, and cannot control the false discovery rate, whereas the justified permutation method with better performance in false discovery rate leads to more reliable results.

Keywords

• microarray data
• gene expression
• alzheimer disease
• false discovery rate
• methods