umap not set up to be reproducible in all methods

[alanocallaghan]

See hhcho/densvis#9

You set random_state in some of the runumap methods but not others.

eg you do here

seurat/R/dimensional_reduction.R

Line 1496 in 13b615c

random.state <- sklearn$utils$check_random_state(seed = as.integer(x = seed.use))

but not in the default method

seurat/R/dimensional_reduction.R

Line 1280 in 13b615c

umap.args <- list(

I could reprex this but it's obvious from the code that not setting the random_state means it won't be reproducible across seeds

[timoast]

Thanks for reporting, this should now be fixed on the develop branch