Different genes using same method
1
0
Entering edit mode
Last seen just now
West Lafayette

Hello DESeq2 community,

I am using DESeq2 package to find DEG. I read vignette but could not find answer to my question.

My experiment has three factors. I designed my experiment in two different ways:main effect and interaction effect.

1. design=~Time+Trt+Tissue
2. design=~Time+Trt+Tissue+Time:Trt+Trt:Tissue+Time:Tissue+Time:Trt:Tissue

I thought that this two design would give me same DEGs, if I compare just the main effects. However, I got different results. For example, when I compared time_30Vs_0 in design 1, I got many significant genes. When I did the same thing for design 2, none of the genes were significant. I think it could be because the second design use the reference level of other factors even in main effect but I guess first design does not. So, now my questions are

1. Is it common to get different results when comparing same main effects in these two designs?
2. If i didn't get significant genes in design 2 for the main effects of time, can I still compare the interaction effect of time with other factors such as tissue and treatment?

Code should be placed in three backticks as shown below


# include your problematic code here with any corresponding output
# please also include the results of running the following in an R session

sessionInfo( )

DESeq2 • 46 views
1
Entering edit mode
@mikelove
Last seen 21 hours ago
United States

See the vignette section on interactions. Especially the diagram.

The main effects are not the same in a linear model with interactions and without.

For further questions though you may want to talk through the meaning of terms in a linear model with a local statistician.

0
Entering edit mode

Thank you, Michael.