Filter Expression Matrix Using Spline Models

splineFilter filters an expression matrix based on spline models fitted to time-series data, optionally considering treatment and subject ID.

Usage

splineFilter(
  exprMat,
  subjectID = NULL,
  time,
  df,
  pCut = 0.5,
  ifFDR = FALSE,
  treatment = NULL,
  refTreatment = NULL
)

Arguments

exprMat

A numeric matrix of expression data, where rows are features and columns are samples.

subjectID

Character. An optional vector of subject IDs corresponding to columns in exprMat. Default is NULL.

time

A numeric vector representing the time points corresponding to columns in exprMat.

df

A numeric value specifying the degrees of freedom for the spline basis.

pCut

A numeric value for the p-value cutoff to filter significant features. Default is 0.05.

ifFDR

A logical value indicating if the false discovery rate (FDR) should be used for filtering. If FALSE, raw p-values are used. Default is FALSE.

treatment

Character. An optional vector of treatment labels corresponding to columns in exprMat. Default is NULL.

refTreatment

Character. An optional reference treatment label for the treatment vector. Default is NULL.

Value

A filtered expression matrix containing only the features that meet the significance criteria.

Details

The function performs the following steps:

1. Converts time points from minutes to hours if both units are present.

2. Removes rows with missing values from the expression matrix.

3. Constructs a design matrix for the spline model, optionally including subject IDs and treatments.

4. Fits a linear model using the design matrix and performs empirical Bayes moderation.

5. Extracts significant features based on the specified p-value or FDR cutoff.