Dear Rory & the Community,

I wanted to ask more like a general question. Can anyone specify how to properly give multiple covariate information in DiffBind and analyse accordingly? I have human samples with multiple metadata, i.e. age, sex, post mortem delay, sample pH, disease status etc. I know I can give them as new columns in sample sheet, but are there proper rules to do that? For example, some columns will be integers, some will be characters, some will be categoricals, etc. And I want to control for all those covariates and perform differential binding analysis between control vs disease. Here is my knowledge on this: "Browsing through different sources on Deseq2, I can understand one can do this with linear modelling, and putting the variable of interest at the last of the model to correct for all the covariates."

my questions: 1. Will DiffBind work like that? I remember from the Vignette, about using Blocking factors, but it seemed to be not exactly as Deseq2 format, code wise. 2. Can DiffBind give the covariate corrected binding scores directly for doing heatmaps? For example, I know in Deseq2, one might need to use some other supplementary tools(limma) to correct for batch effect and then do PCA/heatmap with those corrected counts(ref: from Deseq2 vignette).

As a beginner, can anyone please give me guidance/some reference material where I can look into the code and learn more about it?

P.S. Stay safe in this Corona virus pandemic! And I really really appreciate your help!

The upcoming release of DiffBind will support the ability to include arbitrary designs and complex contrasts, made up of any of the metadata factors Tissue, Factor, Condition, Treatment, Replicate, and Caller. So one could for example have a design such as:

design = "~Replicate + Treatment + Tissue + Condition"

and set any of the design matrix coefficients however you like for testing, e.g.

contrast = c("Condition","Disease","Control")