DESeq2 on GEO dataset

Question

[hariprasadp.iitkgp]

Hi, I am looking to perform Differential Gene expression analysis using R and GEO datasets. But I couldn't find the count matrices in the datasets, it is a raw data and need preprocessing. Can anyone help me how to use DESeq2 libary to preprocess the GEO dataset for Differential gene expression analysis.

[ATpoint]

You have to download the fastq files, see e.g. Fast download of FASTQ files from the European Nucleotide Archive (ENA) and then follow e.g. this Bioconductor RNA-seq workflow https://www.bioconductor.org/packages/devel/workflows/vignettes/rnaseqGene/inst/doc/rnaseqGene.html

[Kevin Blighe]

Which dataset?

[hariprasadp.iitkgp]

GSE3821_series_matrix

Answer 1

[Kevin Blighe]

Okay, that is a microarray, so, you cannot use DESeq2. You should use limma.

To retrieve the already-normalised data, you can simply use:

library(Biobase)
library(GEOquery)
gset <- getGEO("GSE3821", GSEMatrix =TRUE, getGPL=FALSE)
if (length(gset) > 1) idx <- grep("GPL90", attr(gset, "names")) else idx <- 1
gset <- gset[[idx]]

To see the metadata:

pData(gset)

[oludhe]

Hi Kevin,

Just to jump on this thread rather than make a new one, I am having a bit of an issue actually using the eset in DESeq2. I downloaded an expression set, transformed it into a summarizedexperiment object, but getting into a deseq object is giving me errors about the counts. Here is the input code:

 library(Biobase)
library (GEOquery)
library(DESeq2)

gsm <- getGEO('GSE116904') #RNA-seq data

gse <- gsm [[1]]

summ_exp <- makeSummarizedExperimentFromExpressionSet(gse)
ddse <- DESeqDataSet(summ_exp, countData = as.matrix(countData),design = ~ title)

And here is the error:

> ddse <- DESeqDataSet(summ_exp, design = ~ title)
renaming the first element in assays to 'counts'
Error in DESeqDataSet(summ_exp, design = ~title) : 
  some values in assay are not integers

When I specify countData:

ddse <- DESeqDataSet(summ_exp, countData = as.matrix(assay(summ_exp)),design = ~ title)

I get the following error:

Error in DESeqDataSet(summ_exp, countData = as.matrix(assay(summ_exp)),  : 
  unused argument (countData = as.matrix(assay(summ_exp)))

I'm not really sure how to get around this.

[ATpoint]

The raw counts are at the bottom of https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE116904.

Use these and feed them into DESeq2 with DESeqDataSetFromMatrix, follow the manual for this. You are still trying to use microarray functions (getGEO) for RNA-seq data. The RNA-seq data are not available by this function. Check the content of gse it has zero rows so no genes = no counts.

[Kevin Blighe]

Yes, you need raw counts. The data obtained via getGEO may be empty or represent the normalised / transformed data.