Null-free False Discovery Rate Control Using Decoy Permutations
Kun He,Mengjie Li,Yan Fu,Fuzhou Gong,Xiaoming Sun
DOI: https://doi.org/10.48550/arXiv.1804.08222
2021-04-12
Abstract:The traditional approaches to false discovery rate (FDR) control in multiple hypothesis testing are usually based on the null distribution of a test statistic. However, all types of null distributions, including the theoretical, permutation-based and empirical ones, have some inherent drawbacks. For example, the theoretical null might fail because of improper assumptions on the sample distribution. Here, we propose a null distribution-free approach to FDR control for multiple hypothesis testing. This approach, named target-decoy procedure, simply builds on the ordering of tests by some statistic or score, the null distribution of which is not required to be known. Competitive decoy tests are constructed from permutations of original samples and are used to estimate the false target discoveries. We prove that this approach controls the FDR when the statistics are independent between different tests. Simulation demonstrates that it is more stable and powerful than two existing popular approaches. Evaluation is also made on a real dataset.
Methodology,Statistics Theory