README.md

In alexploner/condFDR: Conditional and conjunctional false discovery rate

condFDR

This package allows the calculation of the conditional and conjunctional conditional false discovery rate (cFDR, ccFDR) from two parallel sets of p-values. The typical use case is for these two be p-values across SNPs for two different (but presumably somewhat genetically correlated) phenotypes.

Note: the code is didactic rather than efficient, and has some weird defaults for p-value pre-filtering for ccFDR.

Installation

You can install the current version of cFDR from GitHub with:

# install.packages("devtools")
devtools::install_github("alexploner/condFDR")

Example

This is how you can estimate the conditional FDR for p1, conditioned on p2:

library(condFDR)
data(psynth)
res1 = cFDR(psynth, "p1", "p2", p2_threshold = 1E-5)
head(res1)
head(subset(res1, cFDR < 0.01))

This is how you can estimate the conjuctional conditional FDR for p1 and p2, corresponding to a false discovery rate for SNPs that are associated with both phenotypes:

res2 = ccFDR(psynth, "p1", "p2", p_threshold = 1E-5)
head(res2)
head(subset(res2, ccFDR < 0.01))

Credit

This package is partially based on code from package cfdr, written by James Liley and Chris Wallace, and published under the MIT license.

alexploner/condFDR documentation built on Dec. 31, 2020, 7:43 p.m.