Bug fix: fix repeated adding samples to CountReads.samples during loading so that the mean reference counts for all samples can be calculated accurately. This would slightly affect the parameter minmean in count_analysis and thus affect the inclusion of low-read genes in the analysis.