Contrasts table examples


Simple Contrasts file:
MutA-WT
MutB-WT


The Contrasts file is made on the basis of the Factors file, in this example:
ExperimentStrain
WildType_rep1WT
WildType_rep2WT
Mutant_A_rep1MutA
Mutant_A_rep2MutA
Mutant_B_rep1MutB
Mutant_B_rep2MutB

Example of the Contrast file using multiple factors:

To clarify multiple factors a little, we will consider a hypothetical experiment with three strains and two time points.
The factors are Strain (WT, MutA and MutB) and Time ( T1 and T2)
The Contrast file could be:

WT.T1-WT.T2
WT.T1-MutA.T1
MutA.T1-MutB.T1
MutA.T2-MutB.T2

This means that in the example above in total 4 analyses will be done, so 4 MA plots, 4 TopTables, 4 VolcanoPlots etc. will be generated

The Contrasts file is made on the basis of the Factors file, in this example:

ExperimentStrain Time
WildType_rep1WTT1
WildType_rep2WTT1
WildType_rep1WTT2
WildType_rep2WTT2
Mutant_A_rep1MutAT1
Mutant_A_rep2MutAT1
Mutant_A_rep1MutAT2
Mutant_A_rep2MutAT2
Mutant_B_rep1MutBT1
Mutant_B_rep2MutBT1
Mutant_B_rep1MutBT2
Mutant_B_rep2MutBT2

To decide which contrasts should be made, we need to be clear what are our scientific hypotheses.
In other words, what are we seeking to show?
First, suppose that we wish to find genes that are always higher in Mutants than in Wild Type at T1 and T2.
Then we will need to form the four contrasts WT.T1-MutA.T1, WT.T1-MutB.T1, WT.T2-MutA.T2, WT.T2-MutB.T2, and select genes
that are significantly up for all four contrasts.
Or suppose we wish to establish that the difference between T1 and T2 is small compared to the differences between WT and Mutant.
An efficient way to establish this would be to form the contrasts (WT.T1+WT.T2)/2-(MutA.T1+MutA.T2)/2 and (WT.T1+WT.T2)/2-(MutB.T1+MutB.T2)/2.


reference: EdgeR user guide