Good morning everyone,

I am currently carrying out analyses of bulk RNASeq data from 4 different but similar cell populations that were exposed to the same treatment. I would like to assess the effect of the treatment applied to the cells. After consulting with my team and looking at PCA plots, we would like to compare all treated cells against controls, regardless of cell line, but adjusting for this "line effect". In a nutshell, we want the output to be only the effect of the treatment when analyzing all cells together.

My design matrix is obtained by the following command :

```
designMat <- model.matrix(~0+sampleinfo$Treatment+sampleinfo$Line, data=counts)
```

And the resulting matrix displays two columns for "treated" and "control", and only three for "line" (say, A, B, and C), the fourth line type being "implicit" (is that okay ?).

After that, I am carrying out the analysis with the following contrast :

```
testedContrast <- makeContrasts(treatvscontrol= Treated - Control, levels=designMat)
```

Which gives me the a vector like `c(-1, 1, 0, 0, 0)`

.

Hence my question : is the line effect really adjusted this way ? Because if I make my design matrix without the *Line* columns, I get wildly different results (and way less DE genes).

Thank you in advance for your time and consideration.