limma: extracting coefficients and design matrix?
0
0
Entering edit mode
17 months ago
fr ▴ 170

In a given experiment I am modeling ~0 + treatment*time + cell_type, for which I have treatment (Control and Treated), time (T1, T2, ...) and cell_type (cell1, cell2, ...). Given that I'm looking at the interaction of treatment and time, not all individual factors will be available (e.g. treated and t1) since these are given as linear combinations of the interactions (treated:t1).

When I do:

fit <- lmFit(df, design)


I can then extract the coefficients with fit\$coefficients:

        treatmentTreated        timeT2          timeT3    ...    treatmentTreated.timeT2
gene1   -1.34                   2.34            1.23      ...    3.45
gene2   -2.44                   3.12            0.23      ...    9.12
gene3   -0.37                   1.59            0.76      ...    0.43
...


Now the issue is that I'd like to extract all coefficients but I do not have those for treatmentControl, or timeT1, as those will be given by a linear combination of the other columns (or rather, the inverse). I'm wondering how I can extract those given that I have no intercept? What am I missing in the interpretation of this? Thanks in advance

limma glm • 422 views