Has anyone tried using the combinations argument in hashedDrops? Unfortunately, I cannot fully understand the explanation in the help docs:(

Thanks in advance!

MIkhael

Let's have a look at what ?DropletUtils says:

An integer matrix specifying valid _combinations_ of HTOs. Each row corresponds to a single sample and specifies the indices of rows in ‘x’ corresponding to the HTOs used to label that sample.

So let's say we have a matrix like:

x <- matrix(rpois(1000, 10), ncol=100) # made up
rownames(x) <- LETTERS[1:10]
x[,1:5]
##   [,1] [,2] [,3] [,4] [,5]
## A    9    9   16    8    9
## B   12   15    7   12    9
## C   10    7   14   13   16
## D   13    5   13    7    5
## E    8    7   13   10    6
## F   10   10   12    7    7
## G   12   15    9    6    8
## H    9   13   12   11   11
## I    5   10   11    6    4
## J    9    7   12   11    6

I might then have this combinations:

combinations <- rbind(
    Sample1 = c(1, 5), # Sample 1 is marked by A and E
    Sample2 = c(1, 4), # Sample 2 is marked by A and D
    Sample3 = c(2, 3)  # Sample 3 is marked by B and C
)

and so on. This particular example uses combinations of 2, but you could also imagine higher-order combinations if your experimental design was more ambitious. I do recall there being some constraints though - something like the number of HTOs used in each sample must be less than half the total number of HTOs.