When running hooke the lowly abundant cell types are coming up as the most significant and have the largest delta abundance. Some of these only have 1 or 2 scaled counts and when plotted look like they are below the limit of detection. An issue may be that the standard error is not being computed correctly.