DESeq2 collinearity issue

[chammer]

Hi everybody,

I have been trying to use DESeq2 for analyzing RNA-seq data, and ran into a problem.

The conditions table looks like this:

sample	donor	virus	vpu
DonorA1_01	A1	none	mock
DonorA1_02	A1	CH293	wt
DonorA1_03	A1	CH293	stop
DonorA1_04	A1	CH293	R50K
DonorA1_05	A1	CH293	teth_count
DonorA1_06	A1	CH077	wt
DonorA1_07	A1	CH077	stop
DonorA1_08	A1	CH077	R50K
DonorA1_09	A1	CH077	teth_count
DonorA1_10	A1	STC01	wt
DonorA1_11	A1	STC01	stop
DonorA1_12	A1	STC01	R50K
DonorA1_13	A1	STC01	teth_count
DonorX_01	X	none	mock
DonorX_02	X	CH293	wt
DonorX_03	X	CH293	stop
DonorX_04	X	CH293	R50K
DonorX_05	X	CH293	teth_count
DonorX_06	X	CH077	wt
DonorX_07	X	CH077	stop
DonorX_08	X	CH077	R50K
DonorX_09	X	CH077	teth_count
DonorX_10	X	STC01	wt
DonorX_11	X	STC01	stop
DonorX_12	X	STC01	R50K
DonorX_13	X	STC01	teth_count
DonorY_01	Y	none	mock
DonorY_02	Y	CH293	wt
DonorY_03	Y	CH293	stop
DonorY_04	Y	CH293	R50K
DonorY_05	Y	CH293	teth_count
DonorY_06	Y	CH077	wt
DonorY_07	Y	CH077	stop
DonorY_08	Y	CH077	R50K
DonorY_09	Y	CH077	teth_count
DonorY_10	Y	STC01	wt
DonorY_11	Y	STC01	stop
DonorY_12	Y	STC01	R50K
DonorY_13	Y	STC01	teth_count
DonorZ_01	Z	none	mock
DonorZ_02	Z	CH293	wt
DonorZ_03	Z	CH293	stop
DonorZ_04	Z	CH293	R50K
DonorZ_05	Z	CH293	teth_count
DonorZ_06	Z	CH077	wt
DonorZ_07	Z	CH077	stop
DonorZ_08	Z	CH077	R50K
DonorZ_09	Z	CH077	teth_count
DonorZ_10	Z	STC01	wt
DonorZ_11	Z	STC01	stop
DonorZ_12	Z	STC01	R50K
DonorZ_13	Z	STC01	teth_count

When I specify the model for differential expression analysis as dds <- DESeqDataSetFromTximport(txi, samples, ~vpu+donor+virus), I get an error message:

Error in checkFullRank(modelMatrix) :
the model matrix is not full rank, so the model cannot be fit as specified.
One or more variables or interaction terms in the design formula are linear
combinations of the others and must be removed.

Same problem if the model is “vpu + virus” only.

I understand that this is because of collinearity among the variables but I am not sure how to resolve the issue.

Any help would be highly appreciated!

Thanks!
Chris

[dpryan]

The problem is that "virus==none" is the same as "vpu==mock".

[chammer]

Thanks a lot!
With respect to the most important comparisons, which is VPU "wt" vs. "stop", leaving out the non-infected samples works.
But if I want to compare non-infected vs. infected samples, how could this be resolved? Using an additional dummy variable?
How would that look like?

Thanks!
Chris

[dpryan]

My guess (since I haven't a clue about the background to your experiment) is that you want vpu="Wt" for the mock infected samples. The combination of vpu and virus would then be distinct.

*Edit*: Alternatively, set the virus="none" samples to what they're mock infected with (e.g., CH293), which I suspect will be clearer.

[chammer]

Perfect, thanks a lot for your help, Devon!

Best,
Chris