pbayes: Map p-values to posterior probabilities

In stevehoang/pbayes: Convert p-values into Bayesian posterior probabilities

Map p-values to posterior probabilities

Description

Given a vector of p-values from independent hypothesis tests, calculate the Bayesian posterior probability of the alternative hypothesis being true. The method implemeted is similar to the method described by Erikson et al., 2010 and Allison et al., 2002.

Usage

pbayes(
  p,
  n_boots = 1000,
  alpha = 0.01,
  n_cores = 1,
  subsample = 1,
  max_comp = 4,
  min_null = 0.4,
  opt_method = "L-BFGS-B",
  monotone = TRUE,
  mask_flagpole = TRUE,
  level_pvals = FALSE,
  ...
)

Arguments

p	A numeric vector of p-values
n_boots	A number providing the number of bootstraps used to calculate the convergence statistic.
alpha	A number representing the convergence statistic for fitting the uniform-beta mixture model.
n_cores	A number representing the number of cores to use for the bootstrap calculation.
subsample	The proportion of p to subsample to calculate the beta mixture
max_comp	A number representing the maximum number of non-uniform components to include in the mixture distribution.
min_null	The lower bound on the null distribution mixing fraction.
opt_method	Optimization method. Options are "L-BFGS-B" or "SANN".
monotone	Logical. Only use monotonically-decreasing beta components in the mixture model
mask_flagpole	Logical. During the fitting procedure, mask p-values that are equal to 1. These values can be overrepresented due to p-value calulations on discrete distributions.
level_pvals	Logical. Apply a cubic polynomial fit to a portion of the p-value histogram to level non-uniform trends.
...	Additional parameters to be passed to bbmle::mle2.

Value

A list of 1) the original p-values, 2) the posterior probabilities corresponding to each p-value, and 3) parameters of the fitted mixture model.

References

Allison, D. B., et al. (2002). A mixture model approach for the analysis of microarray gene expression data. Computational Statistics & Data Analysis, 39(1), 1-20. https://doi.org/10.1016/S0167-9473(01)00046-9

Erikson S., et al. (2010). Composite hypothesis testing: an approach built on intersection-union tests and Bayesian posterior probabilities. In Guerra, R., and Goldstein, D. R., (Ed.), Meta-analysis and Combining Information in Genetics and Genomics. (pp. 83-93). Chapman & Hall/CRC.

stevehoang/pbayes documentation built on Feb. 16, 2024, 2:20 p.m.