applying SIMLR to pbmc3k dataset from seurat

[rojinsafavi]

I get a strange clustering solution for pbmc dataset (that I downloaded from seurat dataset)

library(SIMLR)
library(SeuratData)

the dataset:

Dataset | Version | Summary | seurat | species | system | ncells | tech | default.dataset | disk.datasets | other.datasets | notes | Installed | InstalledVersion
pbmc3k.SeuratData | pbmc3k | 3.1.4 | 3k PBMCs from 10X Genomics | 3.1.4 | human | PBMC | 2700 | 10x v1 | raw | NA | pbmc3k.final | NA | TRUE | 3.1.4

head(pbmc3k@meta.data)

    orig.ident  nCount_RNA  nFeature_RNA    seurat_annotations  numeric_annotation
<fct>   <dbl>   <int>   <fct>   <int>
AAACATACAACCAC  pbmc3k  2419    779 Memory CD4 T    1
AAACATTGAGCTAC  pbmc3k  4903    1352    B   6
AAACATTGATCAGC  pbmc3k  3147    1129    Memory CD4 T    1
AAACCGTGCTTCCG  pbmc3k  2639    960 CD14+ Mono  8
AAACCGTGTATGCG  pbmc3k  980 521 NK  2
AAACGCACTGGTAC  pbmc3k  2163    781 Memory CD4 T    1

matrixx = log10(as.matrix(pbmc3k@assays$RNA@data) +1 )
dim(matrixx)
13714 x 2700

set.seed(11111)
example_large_scale = SIMLR_Large_Scale(X = matrixx,
                                        c = 10, kk = 15)
nmi_2 = compare(pbmc3k@meta.data$numeric_annotation, example_large_scale$y$cluster, method="nmi")
nmi_2
0.332772562210941

plot(example_large_scale$ydata,
col = c(topo.colors(10))[pbmc3k@meta.data$numeric_annotation],
xlab = "SIMLR component 1",
ylab = "SIMLR component 2",
pch = 10)

I

[danro9685]

Hi @rojinsafavi,

it is hard for me to give you a complete feedback without looking at the data, but it seems SIMLR is not converging. Does the tool give you any message/warning? Also, have you tried different number of clusters solutions?

Best, Daniele

[rojinsafavi]

Dear @danro9685 ,

Could you share your email with me please so that I can send you the notebook and the data?

Regards, Rojin

[danro9685]

Sure @rojinsafavi, I just sent you an email to the address you have on Github. Best, Daniele