Question: What is the best deseq2 design for this situtation?
0
6 weeks ago by
hrishi27n20
hrishi27n20 wrote:

I am working on analyzing a dataset using DESeq2. The main goal of the analysis is to compare the gene expression profiles between treated and untreated group. Since we also have sex information on the patient, we are also looking at sex-specific differences of the main condition(Treatment). I have added a code snippet below where I was able to compare the treatment effects in males and females separately. I have come across new information on the patients about their heart disease status.

Now my question is what is the best way to write deseq2 design and contrast in order to extract something like, what are the differences between treatment groups for men who also have a heart disease.

     PatientID  Condition   Sex Heart disease
ABC1234    Treated M   Yes
ABC1235    Treated F   No
ABC1236    Treated M   Yes
ABC1237    Treated F   Yes
ABC1238    Treated M   Yes
ABC1239    Untreated   F   No
ABC1240    Untreated   F   Yes
ABC1241     Untreated   M   No
ABC1242     Untreated   F   Yes

Gr <- factor(paste0(Pheno$Condition,Pheno$Sex))
dds <- DESeqDataSetFromMatrix(countData = myFile, colData = Pheno, design= ~Gr)
dds <- DESeq(dds,parallel=TRUE)
resultsNames(dds)
[1] "Intercept"
[2] "Gr_UntreatedM_vs_UntreatedF"
[3] "Gr_TreatedF_vs_UntreatedF"
[4] "Gr_TreatedM_vs_UntreatedF"

Male <- results(dds, contrast=c('Gr','TreatedM','UntreatedM'))
Female <- results(dds, contrast=c('Gr','TreatedF','UntreatedF'))

deseq2 • 100 views
modified 6 weeks ago by Michael Love24k • written 6 weeks ago by hrishi27n20
Answer: What is the best deseq2 design for this situtation?
2
6 weeks ago by
Michael Love24k
United States
Michael Love24k wrote:

How many samples do you have in total? If you have the samples listed above, this is not sufficient to answer, "what are the differences between treatment groups for men who also have a heart disease". If you have many more samples, such that you have sufficient numbers in each combination, you could use the same approach as above but combining also heart disease status.

Thanks Micheal. I have roughly 140 patients.

1

Yes, so then you can combine the three variables in order to look for treatment differences specific to, say, men w/ heart disease.