Hi, I've been encounter errors during the KLD calculation steps, specifically in the KLD2D function. As far as I can tell, the problem occurs when one sample condition (eg DMSO) has an abundance of zero counts, leading to tied ranks and 0-value bandwidths. Should these genes be filtered out? I am hesitant to do this since some of these genes have high counts in the treated samples, suggesting that they are affected by treatment. Are there any other options or workarounds?