Allow more than 1 filter.identity level in 'clonalNetwork()'

[ycl6]

Not sure if this is the behaviour you would like to add to the clonalNetwork function.

This PR allow the filter.identity to accept one or more identity levels, so that users can show relevant interactions of the chosen levels (for example, clusters belonging to the same supercluster).

[ycl6]

Based on the output from filter.identity = NULL and also how weights were calculated in L182 when filter.identity = NULL, the weights should be the average values of the same to/from combinations, or the deduplicated edge.list1 data.frame.

scRepertoire/R/clonalNetwork.R

Lines 179 to 184 in 6a919b9

	for (x in seq_len(nrow(grid))) {
	summary <- c(to = as.vector(grid[x,1]),
	from = as.vector(grid[x,2]),
	weight = num[grid[x,2]]/total.number[as.vector(grid[x,1])])
	edge.list <- rbind(edge.list, summary)
	}

Therefore, I remove unique() from my previous commit and change the way the weights of the edge.list1 is calculated in L209 from sum to mean.

scRepertoire/R/clonalNetwork.R

Lines 207 to 209 in 6a919b9

	edge.list1 <- edge.list %>%
	group_by(to, from) %>%
	summarise(weight = sum(as.numeric(weight)))