Bug: subsample group name parsing and unnecessary dense matrix conversion

[anastasiya-pendragon]

Hello there! Thank you very much for the package!

I wanted to point out two issues that might affect other users.

1. Unnecessary conversion of sparse matrix to dense matrix, which explodes in memory for huge datasets. For me, with 100k peaks and 30k cells, it was impossible to run the method on personal machine. I have used the patched function of create.group.matrices without this line and it worked fine.

epiCHAOS/R/epiCHAOS.R

Line 116 in 34cb72d

counts <- as.matrix(counts)

2. Problem with subsample group name parsing. If the group names do not have underscores, then all groups are just called "group" in the results.

epiCHAOS/R/epiCHAOS.R

Line 271 in 34cb72d

het$state <- sub("-[^_]+$", "", het$state)

I suggest using this line instead.
het$state <- sub("-\\d+$", "", het$state)

Thank you, and I hope it helps.

Best,
Anastasiya