R API of adaFDR

Installation

First, install adafdr and put it in a virtual environment or anaconda environment(recommended). In the installation script, it assumes you use an anaconda environment called py36. You can change it to your virtual environment name in install.R.

Then install the R package for wrapper functions.

Dependency: reticulate

git clone https://github.com/fxia22/RadaFDR.git
Rscript ./RadaFDR/install.R

Or install from github:

install.packages("devtools")
library(devtools)
install_github("fxia22/RadaFDR")

Functions

The R API contains 3 functions and some data loaders. They produce the same results as their python counterparts.

adafdr_test: function(p_input, x_input, K=5, alpha=0.1, n_itr=1500, qt_norm=TRUE, verbose=FALSE,
                      output_folder=NULL, random_state=0L, single_core=TRUE, fast_mode=TRUE,
                      n_full=NULL, covariate_type=NULL, h=NULL)
adafdr_retest: function(res_adafdr, alpha=0.1, n_full=NULL, output_folder=NULL)
                        
adafdr_explore: function(p_input, x_input, alpha=0.1, output_folder=NULL, n_full=NULL,
                         covariate_type=NULL, h=NULL, vis_dim=NULL)

adafdr_test runs AdaFDR as described in the paper, adafdr_retest retest a set of hypothesis with a different threshold. adafdr_explore visualize the features and save it to a folder.

For input format:

p_input is (n,) matrix.
x_input is (n, d) matrix.
res_adafdr is the output of 'adafdr_test'.

Running

See vignettes/scripts/adafdr_example_airway.R for an example.