Question: DESeq2 "Not full rank" Error
0
gravatar for DakJackson
4 weeks ago by
DakJackson0
DakJackson0 wrote:

Hello, I am trying to use DESeq2 to analyze RNA-seq data for a project on sex-change but I keep getting a 'Model matrix not full rank' ERROR. In this study I have three different species, three different sexes and two different tissue types. I am trying to determine genes differentially expressed between species, between sexes of the same species and between tissues of the same species. Here is a portion of my sampling design (I have a total of 40 samples ESC01-ESC39):

SampleName Species Sex Tissue

ESC39 CSTR Female Gonad

ESC38 CSTR Female Gonad

ESC37 CSTR Female Gonad

ESC32 SSUB Herm Gonad

ESC31 SSUB Herm Gonad

ESC30 SSUB Herm Gonad

ESC29 SSUB Herm Brain

ESC28 SSUB Herm Brain

ESC27 SSUB Herm Brain

ESC24 DFOR Herm Gonad

ESC23 DFOR Herm Gonad

ESC21 DFOR Herm Gonad

ESC20 DFOR Herm Brain

ESC19 DFOR Herm Brain

ESC18 DFOR Herm Brain

ESC15 CSTR Female Brain

ESC14 CSTR Female Brain

ESC13 CSTR Female Brain

ESC10 CSTR Male Gonad

ESC09 CSTR Male Gonad

ESC08 CSTR Male Gonad

ESC05 CSTR Male Brain

ESC04 CSTR Male Brain

ESC03 CSTR Male Brain

I previously tried "design = ~ Species + Species:Sex + Species:Tissue" but am unsure how to properly format my model matrix. Any help would be appreciated!

deseq2 • 86 views
ADD COMMENTlink modified 4 weeks ago by swbarnes2240 • written 4 weeks ago by DakJackson0
Answer: DESeq2 "Not full rank" Error
0
gravatar for Michael Love
4 weeks ago by
Michael Love24k
United States
Michael Love24k wrote:

Can you take a look at the section of the vignette that the message points you to? Do you have any particular questions with respect to the information in that vignette section?

ADD COMMENTlink written 4 weeks ago by Michael Love24k

I have read through the vignette and explored the examples available, but am unsure which of my variables need to be nested with which and wether or not I have to make an additional column.

ADD REPLYlink written 4 weeks ago by DakJackson0

I'd recommend discussing with a statistician about how you may want to approach this data. There isn't one single "correct" design but in fact many depending on your scientific assumptions, and whether you assume certain factors modify the effects of other factors. It's more of a conversation to have with a statistical collaborator, as opposed to a question about how to run DESeq2. As the other Answerer pointed out, you can't have a sex specific effect for each species, because some species only have one sex.

ADD REPLYlink written 4 weeks ago by Michael Love24k
Answer: DESeq2 "Not full rank" Error
0
gravatar for swbarnes2
4 weeks ago by
swbarnes2240
swbarnes2240 wrote:

That design will likely not work, and is likely not what you want anyway. (Species is confounded with sex, because you didn't get all the same sexes for all the species, so you can't do that interaction)

To contrast a subset of samples against a subset of samples, make a new column like the vignette shows

http://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#interactions

And contrast one against another.

ADD COMMENTlink written 4 weeks ago by swbarnes2240
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: 213 users visited in the last hour