Set data method does support only TRUE/FALSE annotations with list of names provided

[denklewer]

Current version does not support sample annotations with other values than boolean.

The code

H <- matrix(runif(K * N, min = 0, max = 1), nrow = K )  
W <- t(matrix(runif(K * M, min = 0, max = 10), nrow = K ) )
rownames(W) <- paste("gene", c(1:dim(W)[[1]]))
colnames(H) <- paste( "sample", c(1:dim(H)[[2]]))

# specify pure samples
H[,1] <-  c(1, 0, 0, 0)
H[,2] <-  c(0, 1, 0, 0)
H[,3] <-  c(0, 0, 1, 0)
H[,4] <-  c(0, 0, 1, 0)
H[,5] <-  c(0.2, 0.3, 0.2, 0.3)
V <- W %*% H
sample_annotation <- list(
  sample_name = colnames(V),
  proportions = unlist(lapply(colnames(H), function(sample_name) {
    return(paste((round(100*H[, sample_name]/sum(H[, sample_name]), digits=0)), collapse="|" ))
  })),
  original_corner = unlist(lapply(colnames(H), function(sample_name) { 
    sum((H[, sample_name]/sum(H[, sample_name])) == 1) == 1}))
)
dso <-  DualSimplexSolver$new()
dso$set_data(V, sinkhorn_iterations = 20, sample_anno_lists = sample_annotation)

will cast all annotations provided to TRUE/FALSE

The workaround is to set it afterwards manually.

sample_annotation <-  as.data.frame(sample_annotation)
current_anno <- pData(dso$get_data())
current_anno[,colnames(sample_annotation)] <-  sample_annotation
dso$st$data <- set_anno(anno=current_anno, eset=dso$get_data(), genes = F)

However it would be great to have the ability to provide it in advance

Expected picture is