null_estimation {HDMT} | R Documentation |
A function to estimate the proportions of the three component nulls
Description
This is a function to estimate the proportions of the three component nulls involved in mediation testing. We developed a three component-mixture model method to estimate the proportions of nulls and provide much more accurate control of the family-wise error rate (FWER) and the false discovery rate (FDR), when compared to the standard approach using the uniform null distribution.
Usage
null_estimation(input_pvalues)
Arguments
input_pvalues |
A matrix contains two columns of p-values for candidate mediators. Column 1 is the p-value of testing if the exposure is associated with the candidate mediator (alpha!=0). Column 2 is the p-value of testing if the candidate mediator is associated with the outcome adjusted for the exposure (beta!=0). |
Details
A function to estimate the proportions of the three types of component null hypotheses:
: alpha==0 and beta==0
: alpha==0 and beta!=0
: alpha!=0 and beta==0
The methodology detail can be found in Dai et al (2020).
Value
A list contains five elements.
alpha00 |
A numeric number represents the proportion of null |
alpha01 |
A numeric number represents the proportion of null |
alpha10 |
A numeric number represents the proportion of null |
alpha1 |
A numeric number represents the proportion of null alpha==0 |
alpha2 |
A numeric number represents the proportion of null beta==0 |
Author(s)
James Y. Dai and X. Wang
References
James Y. Dai, Janet L. Stanford, Michael LeBlanc. A multiple-testing procedure for high-dimensional mediation hypotheses, Journal of the American Statistical Association, 2020, DOI: 10.1080/01621459.2020.1765785.
Examples
data(snp_input)
input_pvalues <- snp_input
#To save computing time for illustration, we use 10 percent of rows (p-values)
input_pvalues <- input_pvalues[sample(1:nrow(input_pvalues),
size <- ceiling(nrow(input_pvalues)/10)),]
nullprop <- null_estimation(input_pvalues)