Extremely different results for both EdgeR and DESeq2 analysis
1
0
Entering edit mode
@373f98d7
Last seen 25 days ago
Singapore

Dear all,

Upon comparing my results for the analysis between DESeq2 and EdgeR, I have realized that the 2 results obtained after DEG analysis are extremely different from each other. The thresholds I have set for upregulated genes are Padj < 0.05 and log2FC >0 and log2FC <0 for downregulated genes. I understand that both are different packages and hence can yield different results which have been mentioned in another post. However, there seems to be something more concerning in the 2 results I have obtained.

  1. Out of all the genes reported to be upregulated in both packages; 30% are common between the 2 packages (EdgeR and DESeq2)
  2. Conversely, when comparing genes reported to be upregulated in DESeq2 and reported to be downregulated in EdgeR, there is also a 30% overlap. In this case the 100% constitutes the total number of genes reported to be upregulated in DESeq2 and downregulated in EdgeR.

I am not sure what I could have done wrong. I have also ensured that I releveled both reference conditions to the control.

edgeR DESeq2 • 251 views
ADD COMMENT
0
Entering edit mode
@gordon-smyth
Last seen 1 minute ago
WEHI, Melbourne, Australia

The two analyses don't even appear to be of the same data:

  1. Different count data files
  2. Different column names
  3. Different meta data information
  4. Different DE comparisons (Timepoint 2 vs Control for DESeq2 and Timepoint 1 vs Control for edgeR)
ADD COMMENT
0
Entering edit mode

Thanks for replying! I have these 2 codes in different R scripts, therefore I named them a little differently. The different variable names refer to the same count file. The column names are a little different since I assigned the groups differently in EdgeR and DESeq2. I have used the meta data file for DESeq2 and defined the groups as 1: Control, 2: timepoint 2, 3: timepoint 1 in EdgeR.

I have used EdgeR to find 2 different timepoints, the timepoint1 vs control was the last comparison I made.

Sorry for the confusion caused!

ADD REPLY
0
Entering edit mode

No one here will be able to help you if you make so many arbitrary and unnecessary changes between the two scripts. If you want feedback, you need to demonstrate that you are feeding the same counts, meta data, design matrix and contrast to both packages.

ADD REPLY
0
Entering edit mode

Sure, sorry for the confusion. I will standardize everything before updating the post again.

ADD REPLY

Login before adding your answer.

Traffic: 260 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