• Corpus ID: 211818322

Accurate $p$-Value Calculation for Generalized Fisher's Combination Tests Under Dependence

@article{Zhang2020AccurateC,
  title={Accurate \$p\$-Value Calculation for Generalized Fisher's Combination Tests Under Dependence},
  author={Hong Zhang and Zheyang Wu},
  journal={arXiv: Methodology},
  year={2020}
}
Combining dependent tests of significance has broad applications but the $p$-value calculation is challenging. Current moment-matching methods (e.g., Brown's approximation) for Fisher's combination test tend to significantly inflate the type I error rate at the level less than 0.05. It could lead to significant false discoveries in big data analyses. This paper provides several more accurate and computationally efficient $p$-value calculation methods for a general family of Fisher type… 
H-MAGMA, inheriting a shaky statistical foundation, yields excess false positives
TLDR
The ‘snp-wise mean model’ of Multi-marker Analysis of GenoMic Annotation is often used to perform gene-level testing for association with disease and other phenotypes, but this methodology is unsound, with implications for H-MAGMA results published in Nature Neuroscience regarding genes associated with psychiatric disorders.

References

SHOWING 1-10 OF 54 REFERENCES
TFisher: A powerful truncation and weighting procedure for combining $p$-values
TLDR
This paper extends the classic Fisher’s combination method to a unified family of statistics, called TFisher, which allows a general truncation-andweighting scheme of input p-values, and compares the power of statistics within TFisher family as well as some rare-signal-optimal tests.
Generalized Goodness-Of-Fit Tests for Correlated Data
TLDR
A testing strategy called the digGOF, which combines a double-adaptation procedure (i.e., adapting to both the statistic's formula and the truncation scheme of the input $p-values) and the IT within the gGOF family, which features efficient computation and robust adaptation to the family-retained advantages for given data.
Fisher's method of combining dependent statistics using generalizations of the gamma distribution with applications to genetic pleiotropic associations.
TLDR
This work proposes to use two generalizations of the gamma distribution: the generalized and the exponentiated GDs, and shows that both generalizations have better control of type I error rates than the GD, which tends to have inflated type II error rates at more extreme tails.
Cauchy Combination Test: A Powerful Test With Analytic p-Value Calculation Under Arbitrary Dependency Structures
  • Yaowu Liu, Jun Xie
  • Computer Science
    Journal of the American Statistical Association
  • 2020
TLDR
It is proved a nonasymptotic result that the tail of the null distribution of the proposed test statistic can be well approximated by a Cauchy distribution under arbitrary dependency structures, making the p-value calculation of this proposed test well suited for analyzing massive data.
Distribution of Fisher's combination statistic when the tests are dependent
Many questions in multivariate analysis involve the situation when the number of variables is greater than the available sample size. Combination test statistics provides one method to deal with this
Combining dependent P-values with an empirical adaptation of Brown's method
TLDR
It is shown that the Empirical Brown's method (EBM) outperforms Fisher's method as well as alternative approaches for combining dependent P-values using both noisy simulated data and gene expression data from The Cancer Genome Atlas.
P -values from permutation and F -tests
A modified generalized Fisher method for combining probabilities from dependent tests
TLDR
Modifications to the Lancaster procedure are proposed by taking the correlation structure among p-values into account, and a novel association between B cell pathways and allograft tolerance is identified.
Combining p‐values in large‐scale genomics experiments
TLDR
To allow a stronger claim about a subset of p-values that is smaller than L, two methods with an explicit truncation are investigated: the rank truncated product method (RTP) that combines the first K-ordered p- Values, and the truncated products method (TPM) that combining p- values that are smaller than a specified threshold.
On the optimally weighted z-test for combining probabilities from independent studies
...
1
2
3
4
5
...