Dear all, I'm a bioinformatichian and I'm using DESeq2 package. I have 2 requests: If I have a Multi-factor design, for ex. I have samples Normal and Treated in 3 different time-series (t0,t1,t2), How can I build the matrix correctly and do the analysis? And what about the bias or batch effect? Generally, I use the all samples together and then with the contrast argument I select the comparison. But since you have integrated the log2 shrinking function, I sometimes find abnormal log2fold-change values. Often the contrasts I want to do are not included in the resultsNames (dds) how can I do? How can I shrink these log2FC value and make them as likely as possible? I hope you can help me as soon as possible. Thank you Giovanna

For time-series experiments, there is a tutorial here: Time course experiments.

Regarding batch effects and how to deal with these, there are questions that have been asked in the online communities, including here on Bioconductor Support. Here are some of my own answers:

• https://www.biostars.org/p/266507/#280157
• https://support.bioconductor.org/p/125386/#125387

So, ideal is to include batch in your design formula in order to control for the batch effect when deriving test statistics, and then use limma::removeBatchEffect() when you want to 'subtract out' (eliminate) the batch effect for clustering, PCA, et cetera.

But since you have integrated the log2 shrinking function, I sometimes find abnormal log2fold-change values.

You will have to expand on this point. Can you define 'abnormal'?

Often the contrasts I want to do are not included in the resultsNames (dds) how can I do? How can I shrink these log2FC value and make them as likely as possible?

You can often obtain the desired coefficient by re-leveling a factor (setting a new reference level), and/or by combining / merging factors via paste() or paste0(). Any pairwise contrast can also be performed via: Contrasts

please also note this informative table on fold-change shrinkage in relation to contrasts and other points: Extended section on shrinkage estimators.