hpgl_voom: A slight modification of limma's voom().

In elsayed-lab/hpgltools: A pile of (hopefully) useful R functions

A slight modification of limma's voom().

Description

Estimate mean-variance relationship between samples and generate 'observational-level weights' in preparation for linear modeling RNAseq data. This particular implementation was primarily scabbed from cbcbSEQ, but changes the mean-variance plot slightly and attempts to handle corner cases where the sample design is confounded by setting the coefficient to 1 for those samples rather than throwing an unhelpful error. Also, the Elist output gets a 'plot' slot which contains the plot rather than just printing it.

Usage

hpgl_voom(
  dataframe,
  model = NULL,
  libsize = NULL,
  normalize.method = "none",
  span = 0.5,
  stupid = FALSE,
  logged = FALSE,
  converted = FALSE,
  ...
)

Arguments

dataframe	Dataframe of sample counts which have been normalized and log transformed.
model	Experimental model defining batches/conditions/etc.
libsize	Size of the libraries (usually provided by edgeR).
normalize.method	Normalization method used in voom().
span	The span used in voom().
stupid	Cheat when the resulting matrix is not solvable?
logged	Is the input data is known to be logged?
converted	Is the input data is known to be cpm converted?
...	Extra arguments are passed to arglist.

Value

EList containing the following information: E = The normalized data weights = The weights of said data design = The resulting design lib.size = The size in pseudocounts of the library plot = A ggplot of the mean/variance trend with a blue loess fit and red trend fit

See Also

[limma::voom()]

Examples

## Not run: 
 funkytown = hpgl_voom(samples, model)

## End(Not run)

elsayed-lab/hpgltools documentation built on May 9, 2024, 5:02 a.m.