getBatchEffects: Get experimental batch effects

In eachanjohnson/concensusGLM: concensusGLM

Description

Estimate experimental batch effects.

Usage

getBatchEffects(x, ...)

## Default S3 method:
getBatchEffects(x, ...)

## S3 method for class 'concensusWorkflow'
getBatchEffects(x, ...)

## S3 method for class 'concensusDataSet'
getBatchEffects(x, grouping = c("compound",
  "concentration", "strain"), ...)

Arguments

x	concensusWorkflow or concensusDataSet.
...	Other arguments.

Details

Columns in the data attribute of the concensusDataSet object which have fewer than 100 unique values and which are not named according to a list of stopwords are assumed to be experimental handling annotations, and will be modeled using a Negative Binomial GLM, using dispersion parameters in the dispersion attribute.

Only the negative control observations will be used for estimating batch effects.

A new column, predicted_null_count, is added to data of concensusDataSet with a prediction of number of counts for every observation, assuming only batch effects (and no real effect of interest).

If the model fitting fails, rather than throw off the whole pipeline, a predicted_null_count value of 1 is used.

Value

concensusWorkflow or concensusDataSet with a new batch_effect_model and a new predicted_null_count column in the data attribute.

See Also

glm

eachanjohnson/concensusGLM documentation built on June 26, 2019, 2:26 a.m.