Pearson Correlation Single cell seurat

Question

[Payal]

Hello,

I am doing my singlecell analysis in Seurat. I want to create Pearson correlation matrix of a set of genes per cluster basis. I want to see how these set of genes are co-expressed in all clusters.

Something like this(Referenced from Deseq2 manual)

I know about Seurat has FeatureScatter but that's only one gene against another. How to do for multiple genes? Also which data slot is appropriate for this kind of analysis?

Thanks

[LChart]

Can you be a bit more specific about how you want this to break down? Do you want one correlation matrix per cluster and to visualize the matrices? Or do you actually want scatterplots like FeatureScatter? And, within the same figure, do you want to look at all pairs of genes within one cluster; or to look at the same pair of genes across multiple clusters?

[Payal]

I want to look at all pairs of genes within one cluster. I want one correlation matrix per cluster for a set of around 90 genes I am interested in. I actually want to create a plot like below.

Example: GeneA to GeneD in row for cluster 1 and again GeneA to D in column for cluster 1, with their correlation values.

[LChart]

Something like this

clusterCorPlot <- function(seurObj, cluster.id, gene.list, assay='RNA', slot='data') {
  gene.dat <- getAssayData(
      subset(seurObj, seurat_cluster == cluster.id), 
       assay=assay, 
       slot=slot)[gene.list,]
  ggcorrplot(as.data.frame(t(gene.dat)))
}

should be a good starting point. (Untested)