search

WWXkenmo / ICAnet

Independent Component Analysis decipher functional modules for better cell clustering and annotation
15 stars 1 forks source link

clusters overlap in UMAP #5

Closed Roger-GOAT closed 2 years ago

Roger-GOAT commented 3 years ago

Hi, me again! I run the ICAnet result in Seurat way. ICAnet <- FindClusters(ICAnet,graph.name = 'IcaNet_TF_snn',algorithm = 4,resolution = 0.6) ICAnet <- RunUMAP(ICAnet,reduction = 'pca',seed.use = 123,graph = 'IcaNet_TF_snn',assay = 'IcaNet_TF',min.dist = 0.01,n.neighbors = 40) DimPlot(ICAnet,label = T,group.by = 'IcaNet_TF_snn_res.0.6',split.by = 'orig.ident') The clusters overlap in 1,2,3,4,8. Can it be solved? Thanks. image

WWXkenmo commented 3 years ago

Hi! If the clustering results and feature reductions results are inconsistent, try to make sure you using the same graph to perform both reductions and cell clustering. I was wondering if you used the "pca" reductions to construct 'IcaNet_TF_snn'?