#BCconf
An R package that estimates and adjusts for latent factors in high dimensional biological data using the bias-corrected method described in https://arxiv.org/abs/1801.00865v2

##Model description
The model for the p x n data matrix Y (p = num. of genes, CpG sites, etc., n = sample size) is Y = BX' + LC' + E, where gth row of E has mean 0 and covariance sigma^2_g * I_n. X is an n x d model matrix that contains all observed covariates to be included in the model and C are unobserved factors, like cell composition and/or technical factors. The goal is the estimate the columns of B corresponding to the columns of interest in X (i.e. disease status). The package also estimates C, which can be used in downstream analyses.

##Package functions
Correction: Estimate C and the columns of interest in B (i.e. the effects due to the covariates of interest), as well as corresponding test statistics and p-values. This is the main function.
est.conf.num: Estimate the number of latent factors, i.e. the number of columns in C, using CATE (see https://cran.r-project.org/web/packages/cate/cate.pdf)
ChangeMethod: Change the confounder correction method.
Sig.Omega: Compute the p-value for the null hypothesis that the design matrix X is not correlated with the latent factors C. All factors should be left in the final model even if this is not significant.  

##Installation instructions:
When installing, users may see the error message "Skipping 1 packages not available: sva". Apparently this is a new devtools feature, as it now attempts to update packages that BCconf is dependent on (the package cate depends on sva). We recommend installing sva before install BCconf to avoid this message, which can be done by typing the following into the R console:
source("https://bioconductor.org/biocLite.R")
biocLite("sva")

To install this R-package, first install "irlba", "cate" and "devtools" from CRAN if you do not already have them. Then type the following into the R-console:
devtools::install_github("chrismckennan/BCconf/BCconf")

##Code to reproduce results from our paper
We have included code and instructions in the Rmarkdown file Simulations_181212.Rmd to reproduce the simulation results from the most recent version of our paper. Make sure to download the two .R files in the "Simulations" directory before running Simulations_181212.Rmd.