Biased results of DESeq2 LRT (Group name order)
1
0
Entering edit mode
@minyoungcho93-22184
Last seen 4.2 years ago

Hello,

I am using DESeq2 to perform LRT analysis. I have 4 conditions, and each group named Group1, Group2, Group3, and Group4. Each group has 80 to 100 samples. I use only p-adj value to determine DEGs.

Whenever I use DESeq2 with LRT function, the last group (like group4) has distinctive characteristics. When I changed the group name (group3 to group4, group4 to group3), DEGs were changed. (common DEGs were only 1/4 or 1/5)

Should I use other values to determine DEGs except for p-adjust value? Or should I change the design or other codes?

DESeq2Table <- DESeqDataSetFromMatrix(countData = countData, colData = colData, design = ~ Group_Name) DESeq2Table <- DESeq(DESeq2Table, test="LRT", reduced=~1)

DESeq2Table <- estimateSizeFactors(DESeq2Table) DESeq2Table <- estimateDispersions(DESeq2Table)

DESeq2Table <- nbinomWaldTest(DESeq2Table) DESeq2Res <- results(DESeq2Table, pAdjustMethod = "BH")

Thanks,

deseq2 • 397 views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 9 hours ago
United States

Above you have code for the Wald test, is this on purpose? Do you know that you’re doing pairwise comparisons?

You should remove these lines:

DESeq2Table <- estimateSizeFactors(DESeq2Table) DESeq2Table <- estimateDispersions(DESeq2Table) DESeq2Table <- nbinomWaldTest(DESeq2Table)

ADD COMMENT
0
Entering edit mode

Thanks for the reply! I'll try it again.

ADD REPLY

Login before adding your answer.

Traffic: 849 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6