DESeq with two crossed factors

[coutellec]

Dear all,

I wish to use DESeq to analyse a RNAseq dataset composed of 16 libraries under a design with two crossed fixed factors (2 levels each) and a random factor (pop). In other words, I xwould like to apply function nbinomGLMTest with the following design:

count~factor1*factor2 +1|pop


Alternatively, if random factors cannot be accounted for, I might apply:
count~factor1 * factor2 + factor2/pop

How should I prepare the datafile for such a design? Is there a possibility to include columns to assign counts to the factors levels?
Should I normalize considering all combinations of the two fixed factors (4 conditions in all)?

Thanks in advance for any help
Marie-Agnès

[Simon Anders]

When calling 'newCountDataSet', pass a data frame as second argument with one column for each factor and one row for each sample (rows in the same order as the columns in the count table, of course). Use this to specify the assignment of factor levels to samples.