Clustering artefacts when processing single-nucleus RNAseq

0

Entering edit mode

3.5 years ago

MikeV ▴ 10

Hello,

I've been downloading and processing publicly available 10x genomics single-cell data. But I've been having a particularly hard time with a single-nuclei AD dataset (GSE174367).

Specifically, in addition to cell-type specific clustering I am seeing hundreds of tiny clusters. I've tried setting the expected doublet rate higher than normal when removing doublets, as well as implementing stricter QC filtering thresholds, but they remain... Even at high cluster resolutions the smaller clusters are often grouped with the larger ones.

enter image description here

Has anyone seen this type of artefact before?

single-cell RNA Seurat UMAP • 913 views

ADD COMMENT • link 3.5 years ago by MikeV ▴ 10

0

Entering edit mode

what tools produces this plot? sometimes the number of clusters is specified as an external parameter

ADD REPLY • link 3.5 years ago by Istvan Albert 102k

0

Entering edit mode

This is using the harmony + Seurat workflow.

o <- Seurat::RunUMAP(o,reduction = "harmony", dims = 1:30)
o <- Seurat::FindNeighbors(o,reduction = "harmony", dims = 1:30)
o <- Seurat::FindClusters(o,resolution = 0.1)

You're right though, I haven't tried tinkering with UMAP/clustering settings. Will give that a go.

ADD REPLY • link 3.5 years ago by MikeV ▴ 10

Login before adding your answer.