Can I use Tobias to make non-pairwise comparisons?

[Rosmaninho]

Hey I have two different cell types across three compartments. Is it possible to use Tobias to determine differential binding between one variable independently of the other (cell types and compartments) (2 variables) as it is with edgeR for RNA-seq data (design matrix)?

[msbentsen]

Hi Rosmaninho,

I just want to make sure I understand you correctly - do you want to make a comparison something like this?

Or only within celltypes / compartments?

[Rosmaninho]

I want to use a linear model to eliminate dependent variables when performing comparisons. For instance, I have cell 1 in 3 compartments and cell 2 also in three compartments. For example, among other things, I want to perform the comparison between cell types without the different compartments confounding that comparison.

https://www.middleprofessor.com/files/applied-biostatistics_bookdown/_book/factorial.html#a-linear-model-with-crossed-factors-estimates-interaction-effects

https://www.r-bloggers.com/2020/09/generalized-linear-models-and-plots-with-edger-advanced-differential-expression-analysis/

[msbentsen]

Ah, I understand now - no, that is unfortunately not possible with TOBIAS. If you run TOBIAS on the individual replicates, it might be possible to calculate the effect afterwards based on the mean footprint scores per TF, but it is not something that I have tried out yet. Sorry! I hope you can find a solution nevertheless.