ruvNorm: Remove unwanted variance (RUV)

In biobenkj/ATACseeker: Utilities for ATACseq QC, differential accessibility, and integration

Description

This function is a wrapper for RUVs from RUVSeq to remove unwanted variance using control genes. Additionally, if the number of factors to remove is unknown, the value of k is estimated using surrogate variable analysis. However, if the number of factors is known a priori, then k can be specified.

Usage

ruvNorm(x, mod, cIdx, k = "estimate", scIdx, round = TRUE, epsilon = 1,
  tolerance = 1e-08, isLog = FALSE, ...)

Arguments

x	A data matrix where loci are rows and samples are columns
mod	A model matrix of variables to fit the data
cIdx	A vector of loci/spike-ins in the data matrix to be used as negative controls (default = all loci)
k	The number of factors of unwanted variation to be estimated/removed from the data (default = estimate)
scIdx	A matrix of sample replicates to be used to estimate/remove unwanted variation
round	Measures are rounded to form pseudo-counts
epsilon	An offset to be added to avoid taking log of 0 (default = 1)
tolerance	Selection tolerance for singular values to be considered positive (default = 1e-8)
isLog	Whether the input matrix is already log-transformed (default = FALSE)
...	Other arguments to be passed to estK

Value

A list object of the samples-by-factors matrix of estimated unwanted variation and the normalized counts (removed unwanted variation)

biobenkj/ATACseeker documentation built on May 7, 2019, 8:36 a.m.