Differential gene expression design using Deseq2

Question

[drajangirija]

I am having RNA-seq gene count data for 20 samples.

M5_1
M5_3
M5_7
M5_8
M5_12
MT7_1
MT7_3
MT7_7
MT7_8
MT7_12
M9_1
M9_3
M9_7
M9_8
M9_12
MT1_1
MT1_3
MT1_7
MT1_8
MT1_12
…

How to create design in Deseq2 for different pairs e.g.,

M5_1 vs MT7_1
M5_3 vs MY7_3 
..
..
..
M9_1 vs MT1_1
M9_3 vs MT1_3
..
..

and

M5_1 vs all M5 MT7_1 vs all MT7 M9_1 vs all M9 MT7_1 vs all MT7.

The study design of these data

Answer 1

[CTLong]

Can you provide more information about the biological question you are trying to answer with the RNA-seq analysis? It is rare to pairwise compare between individual samples. Are you trying to compare between conditions (i.e., M5, MT7, M9, MT1)?

If that is the case, make sure that the metadata contains two columns, columnA contains the names of your sample and columnB contains the corresponding condition. With your counts data and metadata, you can load it into popular RNA-seq packages such as DESeq2 and perform pairwise comparison between conditions by specifying the contrast.

dds <- DESeqDataSetFromMatrix(countData = counts,
                          colData = metadata,
                          design= ~ condition)
dds <- DESeq(dds)
res1 <- results(dds, contrast=c("condition","M5","MT7"))
res2 <- results(dds, contrast=c("condition","M9","MT1"))

[drajangirija]

Thanks for your response. The four IDs represent the sample IDs: two responders and two non-responders. The samples were collected from different sites, designated as Site 1, ID_1, and so on. I want to compare the responders versus non-responders at each site.

[swbarnes2]

Two versus two is not a strongly powered study. You might have been better off with more replicates at one site.