findid-methods: Find the cell class with the highest/lowest average channel

In kieranrcampbell/SpatialStats: Statistics for spatially resolved proteomics data

Description

Given tissue samples with different classes, it may be advantageous to single out a class that corresponds to a certain phenotype. These are easiest found by considering the cell class with the highest or lowest average (mean/median) expression of a given tissue, e.g. in epithelial cells keratin is over-expressed while in stromal cells vimentin is under-expressed.

Apply FindID across an entire SPExp

Usage

FindID(sp, channel, direction = c("higher", "lower"), method = c("mean",
  "median"))

FindIDs(SPE, channel, direction = c("higher", "lower"), method = c("mean",
  "median"))

Arguments

sp	An SPData object
channel	A (partial) channel name
direction	Whether to look for the highest or lowest average
method	The method for average (mean or median)
SPE	An SPExp object

Details

Most normalization routines normalize each class separately. However, this will make each channel N(0,1) within a cell class, so comparison is just noise. This method normalizes the selected channel w.r.t. cell size and concentration, then picks out the highest or lowest average (mean/median, set by method).

channel greps the channel name to find an index containing the string, but make sure that string exists in a channel and only once, otherwise an error will be thrown.

direction If "higher" returns the class with the highest mean/median of the selected channel, while "lower" will return the class with the lowest mean/median.

Value

The class index with the higher average in the channel

A vector of cell classes corresponding to each sample.

Examples

## Not run: 
## Keratin is higher in epithelial cells:
keratin.class <- FindID(sp, "Keratin", direction="higher", method="median")

## End(Not run)

kieranrcampbell/SpatialStats documentation built on May 20, 2019, 9:24 a.m.