DESeq2 comparison two cell types and two genotypes

Question

[yosapol.harn]

Dear All,

I have the samples of two cell types (Embryo and Ovule) with two genotypes (WT and MT), which indicate as condition and genotype below. Could you please suggest how to extract information from these four combinations?

I would like to compare:

1. EmbryoWT vs EmbryoMT
2. OvuleWT vs OvuleMT.
3. EmbryoWT vs OvuleWT
4. EmbryoMT vs OvuleMT

I used the code below with model design = ~ genotype + condition + genotype:condition to compare my samples and try to look for solutions. However, It seems to be the analysis will include the same condition or genotype together which is not our interest.

`dds <- DESeqDataSetFromMatrix(countData = clean_data, colData = meta, design = ~ genotype + condition + genotype:condition)

I highly appreciate your help.

Thanks in advance,

[Kevin Blighe]

Has been cross-posted: https://support.bioconductor.org/p/9136689/

Answer 1

[swbarnes2]

You can use interactions, but don't. Make a new column of tissue_genotype, make that your design.

[yosapol.harn]

Hi swbarbes2

Thank you very much for your comments. Do you mean the interaction of genotype:condition from this model?

    design = ~ genotype + condition + genotype:condition

Could you please suggest how to extract this condition?

1. EmbryoWT vs EmbryoMT
2. OvuleWT vs OvuleMT.
3. EmbryoWT vs OvuleWT

4. EmbryoMT vs OvuleMT

   Best regards,

[Assa Yeroslaviz]

I think what swbarbes2 meant was to create third column concatenating the two other columns. You'll than have this column to make the design from

EmbryoMT
EmbryoMT
EmbryoMT
EmbryoWT
EmbryoWT
EmbryoWT
OvuleMT
OvuleMT
OvuleMT
OvuleWT
OvuleWT
OvuleWT

This you can use to make your analysis with.

To make your pair0wise comparison you can look at the resultNames() command or manually create your pairs, if what you need is not there by looking at the contrasts option

[yosapol.harn]

Hi Assa,

Thank you very much for your suggestion. In this case, which model would you suggest?

design = ~ genotype + condition + genotype:condition or only design = ~ concatenation

Best regards,

[swbarnes2]

The second. With a complex call to results, you can get the first design to answer your questions, but using the simpler design is much harder to mess up, and much easier for you or someone else to follow down the road.

[yosapol.harn]

Dear swbarnes2,

Thank you very much for pointing this out. Based on this model, I can find EmbryoWT vs EmbryoMT and OvuleWT vs OvuleMT.

Would it be possible to find ALL WT (EmbryoWT+OvuleWT) vs ALL MT (EmbryoMT+OvuleMT)**?

Thank you very much for your help

[Assa Yeroslaviz]

Just use the original column in your design with these two factors WT and MT for that.