Seurat: RunTSNE fails if data has duplicates

Created on 24 May 2018 · 4Comments · Source: satijalab/seurat

Hi,

RunTSNE(do.fast=T) fails if data contains duplicates.
error in Rtsne.default(X = as.matrix(x = data.use), dims = dim.embed, : Remove duplicates before running TSNE

that probably happens because default argument in Rtsne is check_duplicates=TRUE - but in huge datasets natural duplicates can occur. Is it possible to add "check_duplicates" to RunTSNE function? Or is it naturally impossible to use fast tSNE with duplicates?

Analysis Question duplicate

Source

azkalot1

Most helpful comment

Hi @azkalot1,

You can pass that argument to the Seurat::RunTSNE() function as follows:

object <- RunTSNE(object = object, check_duplicates = FALSE)

Best,
Leon

leonfodoulian on 24 May 2018

👍2

All 4 comments

Hi @azkalot1,

You can pass that argument to the Seurat::RunTSNE() function as follows:

object <- RunTSNE(object = object, check_duplicates = FALSE)

Best,
Leon

leonfodoulian on 24 May 2018

👍2

Please see #85 and #167

mojaveazure on 24 May 2018

Hi,

I got the same error in Seurat 3.0. But "check_duplicates" does not work in Seurat 3.0. What the name of parameter for "check_duplicates" in Seurat 3.0? I noticed many parameters have different names in 3.0, which I have not find a document on that. Thank you so much!