I had a question regarding MaxEntScores. How normal is it to get negative MaxEntScores? When I take my cufflinks results and threshold via FPKM (but still take the majority of the genome), then histogram the MaxEntScores of the junctions in genes above the threshold, I get a bimodal distribution with half of the junctions (both 5' and 3') having negative MaxEntScores and half the junctions having positive MaxEntScores (as well as negative means). This differs quite a bit from the analysis in fig 3 of Yoshimoto et. al. (http://rnajournal.cshlp.org/content/23/1/47.full), where it appears all the MaxEntScores are positive, even for the full genome analysis. Am I doing something obvious wrong?