single cell analysis: normalize gene expression across samples

Question

[neta.zuckerman]

Hi,

I have results from a 10x genomics single cell experiment. In the experiment I have sequenced 7 different samples, and would like to examine the gene expression / cell type differences between them.

I already ran Cell Ranger "count" function on each of these samples. My question is - can I now compare between the genes expressed in each of these samples? (e.g. compare gene expression in cells identified in clustering analysis as "microglia" in sample 1 vs. cell identified as "microglia" in sample 2?). Or - do I need to normalize the gene expression between the different samples somehow? if so, how? it doesn't seem that using Cell Ranger's "aggr" function is for this purpose, but maybe I'm wrong.

Thanks!

[zizigolu]

I noticed that Seurat R package is very user friendly. May you could do a violin plot for each of samples or whatever, However, Seurat is very easy to understand; For instance for normalization Seurat just divides the raw read counts of each sample on the sum of reads of that sample. Then all of matrix times by 10000, finally getting natural log of Matrix of normalized counts.