cluster with extreme low nCounts

[tu-jihao]

For LMS samples in GSE212527, a public soft tissue sarcoma single dataset, I filtered the seurat object out with nFeature in 200-9000. And then delete the doublets, assuming 7.5% doublet formation rate. After clustered by FindClusters function, I found some clusters (c0, c3) has obvious lower Counts number than others (and also has fewer and inapparent positive markers). Does this clusters indicate poor sequencing? Maybe we should delete this clusters before futher analysis, since it is very hard to annotate them.