How do I load a GEO dataset into Seurat?

Question

[Audra]

I'm very new to single cell clustering, but have been able to get results from Seurat using sample datasets from 10x Genomics and also some datasets that were in H5 format.

I'm looking for scRNA-seq datasets that are specific to traumatic brain injury (TBI). So far I haven't found any datasets that are in a format I am familiar with. One looks promising: GSE101901. I'm not sure how to load this one into Seurat... there are several other GEO TBI-related datasets out there but it's the same problem.

Can you show me how to either 1) find a TBI-related scRNA-seq dataset in a format I am familiar with or 2) successfully create a seurat object with the GSE101901 dataset?

Thank you for your help, I've been stuck on this for some time!

Answer 1

[ATpoint]

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?&acc=GSE101901 has a supplementary file which is a count matrix. It has cells as columns and rows as genes. It seems a bit misshaped because there is no name for the first column so standard approaches such as read.delim will complain. I managed to get this file into R with data.table::fread which then simply calls the first column with the gene names V1. You can then move this column to rownames and then make your Seurat object with it, described in the Seurat docs.

library(data.table)

ftp <- "https://ftp.ncbi.nlm.nih.gov/geo/series/GSE101nnn/GSE101901/suppl/GSE101901_DropSeqTBI.digital_expression.txt.gz"
counts <- data.table::fread(ftp)

rownames(counts) <- counts$V1
counts$V1 <- NULL

library(Seurat)
Seurat::CreateSeuratObject(...read...Seurat...docs)

[Audra]

Thank you so much!

[Mgnew]

I used the similar method for a few GEO databases and they worked very well. However I got something very weird from the following GEO database: https://ftp.ncbi.nlm.nih.gov/geo/samples/GSM5555nnn/GSM5555214/suppl/GSM5555214_Vthunter_0029_count_new_data_20210531.txt.gz

Any help?

Thanks!