kmeans error when running reduceDimension(reduction_method = 'DDRTree')

Question

[mk]

After setting my ordering filter, I run:

colonTumor <- reduceDimension(colonTumor, max_components = 3, norm_method = c("none"), reduction_method = 'DDRTree', verbose=TRUE)

Then I get the following error:

Error in kmeans(t(Z), K, centers = centers) : initial centers are not distinct

Here is some info about the data I have used to create the CellDataSet:

Browse[2]> str(colonTumor@assayData$exprs) Formal class 'dgCMatrix' [package "Matrix"] with 6 slots ..@ i : int [1:9499265] 2 5 14 15 23 24 27 29 30 31 ... ..@ p : int [1:1905] 0 5478 11248 15546 21080 24719 28635 32894 38503 42469 ... ..@ Dim : int [1:2] 19326 1904 ..@ Dimnames:List of 2 .. ..$ : chr [1:19326] "RP11-34P13.7" "FO538757.2" "AP006222.2" "RP4-669L17.10" ... .. ..$ : chr [1:1904] "Edge1_AAACGGGGTACCGTTA" "Edge1_AAACGGGTCTGGCGAC" "Edge1_AAAGATGAGCCTCGTG" "Edge1_AAAGCAAAGACTGTAA" ... ..@ x : num [1:9499265] 4 1 6 6 5 2 1 1 1 4 ... ..@ factors : list()

Answer 1

[Kevin Blighe]

Something is weird about your dataset... it's too 'flat', i.e., lacking in structure or something along these lines.

This error was additionally reported on GitHub and the developer responded: kmeans centers not distinct in new DDRTree version #26.

Please try the suggestions.

Kevin