Question: How to correctly set up the design in DESeq2 for RRPP?
0
gravatar for gmchaput
22 days ago by
gmchaput0
gmchaput0 wrote:

Hi everyone,

I am analyzing differential gene expression for a plant based on 3'cDNA tag libraries generated from total RNA. We are looking at a fungal treatment (n=6) compared to a sterile control as well as condition (Condition W vs Condition D).

I am using DESeq2 to determine DEGs of the plant when under a fungal treatment compared to the sterile control within the condition. Then I want to use RRPP to compare DEGs and determine which DEGs are unique for each treatment and condition.

I have never used RRPP and reading up on it. Currently, when I import my count data from tximport, I set the design as follows and create a DESeq object:

ddsTxi <- DESeqDataSetFromTximport(txi, sampleTable, design = ~condition + fungal_tr + condition:fungal_tr )
keep <- rowSums(counts(ddsTxi)) > 1 
ddsTxi <- ddsTxi[keep,]
ddsTxi$fungal_tr <- relevel(ddsTxi$fungal_tr, ref = "CON")
ddsTxi$condition <- relevel(ddsTxi$condition, ref = "W")
dds_DESeq <- DESeq(ddsTxi)

So my starting questions are:

  1. Can I use the DESeq object and format for RRPP or do I need to transform via VST?
  2. Is my design incorrect for what I want to do with RRPP downstream?

Thank you and let me know what I need to clarify if need be!

rnaseq deseq2 degs rrpp • 112 views
ADD COMMENTlink modified 22 days ago by Michael Love26k • written 22 days ago by gmchaput0
Answer: How to correctly set up the design in DESeq2 for RRPP?
2
gravatar for Michael Love
22 days ago by
Michael Love26k
United States
Michael Love26k wrote:

What is RRPP? Is this a Bioconductor package?

ADD COMMENTlink written 22 days ago by Michael Love26k

Hi Michael, thanks for responding!

Randomization of residuals permutation procedure (RRPP) is an R package developed by Collyer and Adams but is not a Bioconductor package to my knowledge/checking online.

Our plan was to obtain a matrix of DEGs of the plant for each fungal treatment compared to the sterile control within each condition and then use RRPP to compare that data to determine global transcriptome changes between treatment and conditions. I'm worried I shouldn't be setting up the count matrix or design the way I have it for the DESeq2. Instead I should have one count matrix being for condition W and another being for condition D, and then each of these matrices have the design = ~fungal_tr.

The reason behind this concern may be due to mis-understanding the DESeq2 output. With the design I posted in my question my interpretation of the results was the following:

  1. Results with coefficient FungalTreatment1_Control, these are the DEGs of the fungal treatment compared to the sterile control, controlling for condition.
  2. Results with coefficient ConditionD_FungalTreatment1, these are the DEGs of the plant when infected with fungus #1 when in condition D.
  3. Results with coefficient ConditionW_ConditionD, these are the DEGs of the plant between the two conditions, controlling for fungal treatment.

If this is correct, then I think that when I run RRPP I wouldn't be able to look at DEGs of FungalTreatment1W v.s. DEGs FungalTreatment1D, because I don't have the former.

Let me know if I need to clarify anything. Also side note, your blog and Bioconductor RNAseq workflow docs have been super helpful!

ADD REPLYlink modified 21 days ago • written 21 days ago by gmchaput0

Take a look at the interaction section in the vignette. Your interpretation of 1-3 isn’t exactly correct but we have a diagram of the interactions in the section that may help. Otherwise I’d recommend speaking to a statistician to help with interpretation of the model.

ADD REPLYlink written 20 days ago by Michael Love26k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 147 users visited in the last hour