The support.bioconductor.org editor has been updated to markdown! Please see more info at: Tutorial: Updated Support Site Editor

Question: How to create 3 way interactions model for Deseq
0
gravatar for sharmila.ahmad
11 days ago by
sharmila.ahmad0 wrote:

Hi good day,

This is my first time working on differential gene expression analysis using DeSeq2. I am having a problem to construct my model matrix for DeSeq2 since I have a very complex experimental design (3 factors design) with the following information;

ID Ewed Lambd Sex 1768 HIGH CONV M 1774 HIGH CONV F 1778 HIGH CONV F 1788 HIGH CONV F 1796 HIGH CONV M 1766 LOW CONV F 1772 LOW CONV M 1784 LOW CONV M 1810 LOW CONV F 1816 LOW CONV M 1818 LOW CONV F 1822 LOW CONV M 1826 LOW CONV F 1828 LOW CONV F 1764 NORM CONV F 1770 NORM CONV F 1782 NORM CONV M 1792 NORM CONV M 1802 NORM CONV F 1804 NORM CONV F 1755 HIGH HCHF M 1767 HIGH HCHF F 1777 HIGH HCHF F 1787 HIGH HCHF M 1799 HIGH HCHF M 1823 HIGH HCHF F 1759 LOW HCHF M 1775 LOW HCHF F 1783 LOW HCHF M 1785 LOW HCHF M 1807 LOW HCHF F 1821 LOW HCHF M 1827 LOW HCHF F 1761 NORM HCHF F 1763 NORM HCHF F 1769 NORM HCHF M 1789 NORM HCHF M

I would like to construct the model matrix for three way interaction effects (Ewed:Lambd:Sex) but could not do so in which I got this error "the model matrix is not full rank".

Ewed: Maternal diet Lambd: offspring diet

Could anyone suggest me what can or should I do? I read the DeSeq2 vigenette about this kind of problem but could not really understand it.

Your help is much appreciated.

Best, Mila

deseq2 • 51 views
ADD COMMENTlink modified 11 days ago • written 11 days ago by sharmila.ahmad0

Hi Micheal,

Thank you for your feedback.

I have tried to consult with someone that have knowledge in this matter but could not find any solution to that (for now)

I have also tried to run this following model matrix but got error

design <- model.matrix(~ -1 + per$Lambbw6 + per$Birthw + per$Lambbw2.5, per)

ddstissue <- DESeqDataSetFromHTSeqCount(sampleTable=per, directory=pathht, design = ~ Ewed + Lambd + Sex ) ddstissue

#

factor levels were dropped which had no samples

dds_tissue <- DESeqDataSetFromHTSeqCount(sampleTable=per, + directory=pathht, + design = ~ Ewed + Lambd + Sex ) factor levels were dropped which had no samples

ADD REPLYlink modified 11 days ago • written 11 days ago by sharmila.ahmad0

That is not an error, it is just a helpful message that there were factor levels which had no samples, and so these levels were dropped for you (otherwise the model would be impossible to fit).

ADD REPLYlink written 11 days ago by Michael Love21k
Answer: How to create 3 way interactions model for Deseq
1
gravatar for Michael Love
11 days ago by
Michael Love21k
United States
Michael Love21k wrote:

For users with complex designs involving many interactions, and who are not very familiar with the terms in R’s linear models, I recommend to collaborate with a statistician. I don’t have much time this semester to answer questions that go beyond the DESeq2 software and are more statistical consulting.

ADD COMMENTlink modified 11 days ago • written 11 days ago by Michael Love21k
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: 196 users visited in the last hour