Hello Rory,

Thanks for the Diffbind package, it makes DB analysis very easy.

However, I have the following question: How does DiffBind handle Input samples and how does it compare condition B with condition A? For example, I have a experiment design like this:

Condition    A    B
IP           2     2
IN           2     2

The number 2 indicates 2 biological replicates for each cell.

When DiffBind calls DESeq2, what contrast does it use internally to calculate LFC and FDR between condition A and B?

1. A_IP  - B_IP
2. (A_IP - A_IN) - (B_IP - B_IN) 
3. sth. else?

Thank you!

Ray

The Input reads are not used in the DESeq2 modelling itself. By default, these reads are subtracted from the ChIP reads prior to modelling, you can control this using the bSubControl parameter when calling dba.analyze().