filterConfounders: Filters out the effect of known confounders. This is done by...

In RCM: Fit row-column association models with the negative binomial distribution for the microbiome

Description

Filters out the effect of known confounders. This is done by fitting interactions of every taxon with the levels of the confounders. It returns a modified offset matrix for the remainder of the fitting procedure.

Usage

filterConfounders(
  muMarg,
  confMat,
  X,
  thetas,
  p,
  n,
  nleqslv.control,
  trended.dispersion,
  tol = 0.001,
  maxIt = 20,
  allowMissingness,
  naId
)

Arguments

muMarg	a nxp matrix, the current offset
confMat	a nxt confounder matrix
X	the nxp data matrix
thetas	a vector of length p with the current dispersion estimates
p	an integer, the number of columns of X
n	an integer, the number of rows of X
nleqslv.control	see nleqslv()
trended.dispersion	a vector of length p with trended dispersion estimates
tol	a scalar, the convergence tolerance
maxIt	maximum number of iterations
allowMissingness	A boolean, are missing values present
naId	The numeric index of the missing values in X Fits the negative binomial mean parameters and overdispersion parameters iteratively. Convergence is determined based on the L2-norm of the absolute change of mean parameters

Value

a list with components:

thetas	new theta estimates
NB_params	The estimated parameters of the interaction terms

RCM documentation built on Nov. 8, 2020, 5:22 p.m.