mean_by_bioreplicate: An attempt to address a troubling question when working with...
In elsayed-lab/hpgltools: A pile of (hopefully) useful R functions

mean_by_bioreplicate

R Documentation

An attempt to address a troubling question when working with DIA data.

Description

My biggest concern when treating DIA data in a RNASeqish manner is the fact that if a given peptide is not identified, that is not the same thing as stating that it was not translated. It is somewhat reminiscent of the often mocked and repeated Donald Rumsfeld statement regarding known unknowns vs. unknown unknowns. Thus, in an RNASeq experiment, if one sees a zero, one may assume that transcript was not transcribed, it may be assumed to be a known zero(unknown). In contrast, if the same thing happens in a DIA data set, that represents an unknown unknown. Perhaps it was not translated, and perhaps it was not identified.

Usage

mean_by_bioreplicate(expt, fact = "bioreplicate", fun = "mean")

Arguments

`expt`	Starting expressionset to mangle.
`fact`	Metadata factor to use when taking the mean of biological replicates.
`fun`	Assumed to be mean, but one might want median.

Details

This function therefore does the following: 1. Backfill all 0s in the matrix to NA. 2. Performs a mean across all samples which are known technical replicates of the same biological replicate. This mean is performed using na.rm = TRUE. Thus the entries which used to be 0 should no longer affect the result. 3. Recreate the expressionset with the modified set of samples.