QV depends on coverage, which makes sense because higher coverage of a genomic position leads to higher confidence that the IPD ratio is significantly different from the baseline. Note that this can complicate the comparison of samples : if the coverage levels differ (which is common), a direct comparison of modification QVs and % detected may not be an apples-to- apples comparison. This is particularly problematic if the coverage levels are below 100x, where % detected is very sensitive to coverage. One way to handle this is to always work at >100x coverage, and to compare IPD ratios for each motif when considering: • Identical samples grown under different conditions • Biological replicates. 100 90 80 70 60 50 0 50 100 150 200 250 % Detected Total Coverage 6-mA Detection