Bug in netVisual_heatmap may compute incorrect differences when row/column orders differ between compared networks

[SDelb]

I believe there is a bug in netVisual_heatmap in the visualization.R script.
Around lines 1844-1846, the difference between 2 groups is computed as:

obj1 <- object@net[[comparison[1]]][[measure]]
obj2 <- object@net[[comparison[2]]][[measure]]
net.diff <- obj2-obj1

This assumes that both objects have identical row and column ordering. I found that this is not necessarily the case when comparing CellChat output across conditions, resulting in a heatmap with incorrect values. See below an example of my CellChat results:

Expected behavior:
Before subtraction, the two matrices should be aligned by row and column names.

Actual behavior:
The function directly subtracts matrices without checking ordering of rows and columns.

Suggested fix:

if (! all(rownames(obj1) %in% rownames(obj2))) {
stop("obj1 and obj2 do not contain the same row names (senders).")
}

if (! all(colnames(obj1) %in% colnames(obj2))) {
stop("obj1 and obj2 do not contain the same column names (receivers).")
}

obj2 <- obj2[rownames(obj1), colnames(obj1)]

net.diff <- obj2 - obj1