lmms.filter.lines: Filter Linear Profiles from Linear Mixed Model output

In abodein/timeOmics: Time-Course Multi-Omics data integration

Filter Linear Profiles from Linear Mixed Model output

Description

This function filters linear models with highly heterogeneous variability within residues. From an "lmms" output, 2 parameters are tested:

Usage

lmms.filter.lines(
  data,
  lmms.obj,
  time,
  homoskedasticity = TRUE,
  MSE.filter = TRUE,
  homoskedasticity.cutoff = 0.05
)

Arguments

data	a data.frame used in the lmms::lmmSpline command
lmms.obj	a lmmspline object
time	a numeric vector containing the sample time point information.
homoskedasticity	a logical whether or not to test for homoscedasticity with the Breusch-Pagan test.
MSE.filter	whether or not to test for low dispersion with a cutoff on the MSE.
homoskedasticity.cutoff	a numeric scalar between 0 and 1, p-value threshold for B-P test.

Details

* homo-sedasticity of the residues with a Breusch-Pagan test * low dispersion with a cutoff on the MSE (mean squared error)

Value

a list containing the following items

filtering.summary	a data.frame with the different tests per features (passed = TRUE, failed = FALSE)
to.keep	features which passed all the tests
filtered	the filtered data.frame

See Also

bptest

Examples

# data and lmms output
data(timeOmics.simdata)
data <- timeOmics.simdata$sim
lmms.output <- timeOmics.simdata$lmms.output
time <- timeOmics.simdata$time

# filter
filter.res <- lmms.filter.lines(data = data, lmms.obj = lmms.output, time = time)

abodein/timeOmics documentation built on March 22, 2022, 5:25 p.m.