edgeR GLM paired samples analysis - advice required

Question

[Duff]

Hi

I am seeking some clarification on a design matrix for edgeR.

My experimental design is as follows:

2 groups (knock in & knock out, n=7 per group), within each group I have paired samples from the same animal (vector, empty vector).

The contrasts I would like to make are:

• knock in v knock in empty vector
• knock out v knock out empty vector
• empty vector v empty vector

My current design matrix looks like this:

des <- model.matrix(~ manip + manip:mouse + manip:vector, data=targets)

Where the 'targets' file looks like: <style type="text/css"> .tg {border-collapse:collapse;border-spacing:0;} .tg td{font-family:Arial, sans-serif;font-size:14px;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;} .tg th{font-family:Arial, sans-serif;font-size:14px;font-weight:normal;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;} .tg .tg-yw4l{vertical-align:top} </style>

TargetFile	rat	group_sample	manip	vector
416L_gene_counts.txt	416	1	knock.in	empty.vector
416R_gene_counts.txt	416	1	knock.in	vector
417L_gene_counts.txt	417	2	knock.in	empty.vector
417R_gene_counts.txt	417	2	knock.in	vector
418L_gene_counts.txt	418	3	knock.in	empty.vector
418R_gene_counts.txt	418	3	knock.in	vector
419L_gene_counts.txt	419	4	knock.in	empty.vector
419R_gene_counts.txt	419	4	knock.in	vector
...	...	...	...	...
427L_gene_counts.txt	427	5	knock.out	empty.vector
427R_gene_counts.txt	427	5	knock.out	vector
428L_gene_counts.txt	428	6	knock.out	empty.vector
428R_gene_counts.txt	428	6	knock.out	vector
429L_gene_counts.txt	429	7	knock.out	empty.vector
429R_gene_counts.txt	429	7	knock.out	vector

where manip is knock in or knock out, rat is the specific animal and vector is the active or empty vector.

Using this design matrix I can extract the manip regulated genes with: ki_cont <- makeContrasts('manipknock.in.vectorvector', levels=design) ko_cont <- makeContrasts('manipknock.out.vectorvector', levels=design)

These look good as far as gene lists go.

My questions are:

• Do the above contrasts account for the pair information I have?

• How do I go about extracting the information to compare empty vectors under both manipulations?

Thanks for any help or advice.

Best,

iain

[Carlo Yague]

I can't answer your questions because I don't use edegeR but I think that there is an issue with your design :

~ manip + manip:mouse + manip:vector

should probably be

~ manip + manip*mouse + manip*vector # equivalent of '~manip + mouse + manip:mouse + vector + manip:vector'

[Duff]

Thanks Carlo - I've been following the notation used by the edgeR manual and papers. I've also noticed that 'mouse' should be 'rat' - d'oh. It's fine in my actual design. Too much jumping between species.

best,

iain