dot-sp_costrain: Weigh by spatial constrans
In saezlab/liana: LIANA: a LIgand-receptor ANalysis frAmework

.sp_costrain

R Documentation

Weigh by spatial constrans

Description

Weigh by spatial constrans

Usage

.sp_costrain(lr_res, cell.adj, adjacency = "adjacency")

Arguments

`lr_res`	liana_pipe output prior to joining complexes
`cell.adj`	cell adjacency weights (should be positive)
`adjacency`	name of the column with cell pair adjacency scores

Details

Note that for the case that there are weights from 0-1, the negative values (in e.g. logFC, z-scores) might be counter-logically affected - i.e. they would be brought closer to 0. Thus, by default liana expects weights from 0-1. These are then multiplied for positive values, while negative values are divided.

Alternatively, one could e.g. multiply the weights by a factor (e.g. 10,000), if logFC and Connectome are used. However, this would change some of the assumptions/interpretations of the scores, while consensus ranking will be unaffected.

Also, note that any interactions between cell pairs with an adjacency of 0 will be excluded (this would affect the scores from CytoTalk).

NB! ' single-context, and should not be used when scaling between conditions!!!