do.dual.decon: do.dual.decon - internal

In ERGlass/lrcde.dev: Linear Regression Cell Type-Specific Differential Expression Power Analysis

Description

Performs the core functionality of LRCDE - cell type-specific differential expression detection. This function does the actual linear regression deconvolution per group.

Usage

do.dual.decon(het.sub, cell.props, groups, medCntr = FALSE, stdz = FALSE,
  nonNeg = TRUE)

Arguments

het.sub	the samples (rows) by genes (columns) heterogeneous gene expression matrix. Should contain non-median-centered, non-standardized, positive values. log2-transformation recommended. Required
cell.props	the cell proportion matrix, cell types (rows) by samples (columns). The proportion should be relative, e.g., sum up to ~1 per sample. Required
groups	a vector of 1's and 2's indicating group assignment. Should correspond to the sample order in het.sub and cell.props. Required
medCntr	boolean indicatinng whether to median center difference estimates. Default - FALSE
stdz	boolean indicating whether to standardize difference estimates. Default - FALSE
nonNeg	boolean indicating whether to force cell type-specific expression estimates to be non-negative. Default - TRUE

Details

Here is where the actual group-wise linear regressions are performed. This function is designed to handle a single genomic site at a time; e.g.: a single vector containing het.suberogeneous observations for controls and cases.

Value

Returns a list containing group-wise difference estimates, regression residuals, lm objects per group, and standard errors of cell type-specific expression estimates. Also returns a two item list containing results of Breusch-Pagan test.

Author(s)

Edmund R Glass, [e-mail]

References

https://github.com/ERGlass/lrcde.dev

Examples

## Not run: 
do.dual.decon(het.sub.obs, cell.props, groups, medCntr = FALSE, stdz = FALSE, nonNeg =TRUE)

## End(Not run)

ERGlass/lrcde.dev documentation built on May 6, 2019, 3:09 p.m.