Comparisons of estimators of the number of true null hypotheses and adaptive FDR procedures in multiplicity testing

Many exploratory studies such as microarray experiments require the simultaneous comparison of hundreds or thousands of genes. It is common to see that most genes in many microarray experiments are not expected to be differentially expressed. Under such a setting, a procedure that is designed to control the false discovery rate (FDR) is aimed at identifying as many potential differentially expressed genes as possible. The usual FDR controlling procedure is constructed based on the number of hypotheses. However, it can become very conservative when some of the alternative hypotheses are expected to be true. The power of a controlling procedure can be improved if the number of true null hypotheses (m ₀) instead of the number of hypotheses is incorporated in the procedure [Y. Benjamini and Y. Hochberg, On the adaptive control of the false discovery rate in multiple testing with independent statistics, J. Edu. Behav. Statist. 25(2000), pp. 60–83]. Nevertheless, m ₀ is unknown, and has to be estimated. The objective of this article is to evaluate some existing estimators of m ₀ and discuss the feasibility of these estimators in incorporating into FDR controlling procedures under various experimental settings. The results of simulations can help the investigator to choose an appropriate procedure to meet the requirement of the study.     

Randomized p-values for multiple testing of composite null hypotheses

We are considered with the problem of m simultaneous statistical test problems with composite null hypotheses. Usually, marginal p-values are computed under least favorable parameter configurations (LFCs), thus being over-conservative under non-LFCs. Our proposed randomized p-value leads to a tighter exhaustion of the marginal (local) significance level. In turn, it is stochastically larger than the LFC-based p-value under alternatives. While these distributional properties are typically nonsensical for m  =1, the exhaustion of the local significance level is extremely helpful for cases with m>1$m>1$ in connection with data-adaptive multiple tests as we will demonstrate by considering multiple one-sided tests for Gaussian means.     

A method for modifying multiple testing procedures

Many recent multiple testing papers have provided more efficient and/or robust methodology for control of a particular error rate. However, different multiple testing scenarios call for the control of different error rates. Hence, the procedure possessing the desired optimality and/or robustness properties may not be applicable to the problem at hand. This paper provides a general method for extending any multiple testing procedure to control any error rate, thereby allowing for the procedure possessing the desired properties to be used to control the most relevant error rate. As an example, two popular procedures that were originally designed to control the marginal and positive False Discovery Rate are extended to control the False Discovery Rate and Family-wise Error Rate. It is shown that optimality and/or robustness properties of the original procedure are retained when it is modified using the proposed method.     

A multiple testing protocol for exploratory data analysis and the local misclassification rate

A false discovery rate (FDR) procedure is often employed in exploratory data analysis to determine which among thousands or millions of attributes are worthy of follow-up analysis. However, these methods tend to discover the most statistically significant attributes, which need not be the most worthy of further exploration. This article provides a new FDR-controlling method that allows for the nature of the exploratory analysis to be considered when determining which attributes are discovered. To illustrate, a study in which the objective is to classify discoveries into one of several clusters is considered, and a new FDR method that minimizes the misclassification rate is developed. It is shown analytically and with simulation that the proposed method performs better than competing methods.     

A combined p-value test for multiple hypothesis testing

Tests that combine p-values, such as Fisher's product test, are popular to test the global null hypothesis H₀ that each of n component null hypotheses, H₁,…,H_n, is true versus the alternative that at least one of H₁,…,H_n is false, since they are more powerful than classical multiple tests such as the Bonferroni test and the Simes tests. Recent modifications of Fisher's product test, popular in the analysis of large scale genetic studies include the truncated product method (TPM) of Zaykin et al. (2002), the rank truncated product (RTP) test of Dudbridge and Koeleman (2003) and more recently, a permutation based test—the adaptive rank truncated product (ARTP) method of Yu et al. (2009). The TPM and RTP methods require users' specification of a truncation point. The ARTP method improves the performance of the RTP method by optimizing selection of the truncation point over a set of pre-specified candidate points. In this paper we extend the ARTP by proposing to use all the possible truncation points {1,…,n} as the candidate truncation points. Furthermore, we derive the theoretical probability distribution of the test statistic under the global null hypothesis H₀. Simulations are conducted to compare the performance of the proposed test with the Bonferroni test, the Simes test, the RTP test, and Fisher's product test. The simulation results show that the proposed test has higher power than the Bonferroni test and the Simes test, as well as the RTP method. It is also significantly more powerful than Fisher's product test when the number of truly false hypotheses is small relative to the total number of hypotheses, and has comparable power to Fisher's product test otherwise.