Lack of Contrasts Functionality

[DarioS]

The output of RegParallel is a data.table. Can it be the model fit to allow contrasts? Let's say there is a data that has multiple patients and each patient sample has multiple cell types. It doesn't seem feasible to extract cell type markers by fitting a full model ~cellType + patientID, followed by computing contrasts such as epithelial-(Tcell + Bcell + NKcell)/3 which limma can do.