1
$\begingroup$

I have a RNA seq data which I am trying to identify DEGs. Dataset contains:

Untreated - 2 replicates (time point 1st day)
Treated - 4 replicates (2 replicates at 3rd day & 2 replicates at 7th day )

I merged all the samples and tried to identify DEGs untreated vs treated. The 1st day sample is considered as control (untreated) for the study.

  1. Is it correct to merge all samples and compare. Or should I compare the two time point treated samples with untreated as 1st day vs 3rd day, 1st day vs 7th day and 3rd day vs 7th day separately.

I tried LRT test in DEseq2 to understand the changes in gene expression across time between untreated and treated. The code is given below.

sampleFiles<-grep('SRR',list.files(directory),value=TRUE)
sampleCondition<-c('untreated','untreated','treated','treated','treated','treated')
sampleTable<-data.frame(sampleName=sampleFiles, fileName=sampleFiles, condition=sampleCondition)
ddsHTSeq<-DESeqDataSetFromHTSeqCount(sampleTable=sampleTable, directory=directory, design=~condition)
colData(ddsHTSeq)$condition<-factor(colData(ddsHTSeq)$condition, levels=c('untreated','treated'))
dds <- DESeq(ddsHTSeq, test="LRT", reduced=~1)
resLRT <- results(dds)
resOrdered <- resLRT[order(resLRT$padj),]

After running the code I am getting huge number of DEGs (~15000 genes) as significant DEGs after applying FDR correction cutoff <0.05.

  1. Is there any error in mentioning the condition in the code.

  2. Can you please tell the condition I have specified for LRT test in the code is enough to get the desired output.

  3. Whether LRT test gives huge no of DEGs after analysis as significant DEGs.

$\endgroup$
5
  • 1
    $\begingroup$ Redo this with all 3 groups indicated in condition rather than merging the two latter time-points together. $\endgroup$
    – Devon Ryan
    Commented May 15, 2018 at 10:02
  • $\begingroup$ So all the analysis for the above mentioned conditions(Untreated, Treated T1, Treated T2) must be performed separately? and How can I infer results based on DEG from 3 conditions? $\endgroup$ Commented May 15, 2018 at 11:39
  • 1
    $\begingroup$ All this analysis is really going to tell you is which genes were DE on day 3/7 vs. day 1. In order to determine which gene expression changes are caused by your treatment (as opposed to any number of other factors that might vary between days), you need to compare treated vs. mock treated samples on the same days. $\endgroup$ Commented May 15, 2018 at 12:42
  • $\begingroup$ You can still perform the exact same LRT with 3 groups, it will be more informative than with 2. $\endgroup$
    – Devon Ryan
    Commented May 15, 2018 at 12:44
  • $\begingroup$ " LRT with 3 groups" this step dds <- DESeq(ddsHTSeq, test="LRT", reduced=~1) i have very fundamental doubt. Here it is single condition which is treated/untreatetd under sample condition. LRT is explained in the manual to see the difference between full and reduced model. How does it work here since it compares across the three condition. How do i infer which is being here used as reference here? $\endgroup$
    – kcm
    Commented Mar 29, 2022 at 15:01

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.