Skip to main content
Multiple Comparison Procedures for Neuroimaging Genomewide Association Studies
Biostatistics (2014)
  • Wen-Yu Hua
  • Thomas E Nichols, University of Warwick
  • Debashis Ghosh, University of Colorado Denver

Recent research in neuroimaging has been focusing on assessing associations between genetic variants measured on a genomewide scale and brain imaging phenotypes. Many publications in the area use massively univariate analyses on a genomewide basis for finding single nucleotide polymorphisms that influence brain structure. In this work, we propose using various dimensionalityreduction methods on both brain MRI scans and genomic data, motivated by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) study. We also consider a new multiple testing adjustments inspired from the idea of local false discovery rate of Efron and others (2001). Our proposed procedure is able to find associations between genes and brain regions at a better significance level than in the initial analyses.

  • Genomewide association studies; Distance covariance/correlation; Empirical null estimation; Local false discovery rate; Multiple comparisons; Positive false discovery rate
Publication Date
Publisher Statement
This is an earlier version of the paper; the final one appears in Biostatistics.
Citation Information
Wen-Yu Hua, Thomas E Nichols and Debashis Ghosh. "Multiple Comparison Procedures for Neuroimaging Genomewide Association Studies" Biostatistics (2014)
Available at: