I am working on raw vcf from c.a. 200 samples run in a 4lanes Novaseq. For variant calling I used Lofreq since it is precise and consistent for low vaf snv. Since going low adds noise, I need to manually filter some variants. I am interested in snv observed at different proportions on different lanes. Since from literature it is known that some artifact are lane-specific, I am interested testing whethere some mutations appear prevalently in specific lanes. However, I tested also reference observations and it appeared that there is an umbalancing also for reference positions (for example: L1:260,L2:276,L3:186,L4:200). The point is that I was supposing that both reference and mutations should be balanced or, at least shoul be reference observations. My question is: how is possible that I observed such an unbalancing in many positions?