Fig. 3

Effect of fragment length and GC content on normalization efficiency. The number of sequencing reads of rare (corresponding to AluYb8) and abundant (corresponding to AluYa5) flanks is plotted against fragment length (a) and GC content (b), respectively, in “unnormalized”, “normalization 1” and “normalization 2” datasets. Y axis – number of reads (logarithmic scale). X axis is length of fragments (a) or their GC content (b). Green circles and green crosses indicate Ya5 and Yb8 insertions in “unnormalized” library; orange circles and orange crosses indicate Ya5 and Yb8 insertions in “normalization 1” library; blue circles and blue crosses indicate Ya5 and Yb8 insertions in “normalization 2” library. Trendlines were fit to data using generalized additive models, shaded area indicate confidence interval (CI=0.95) for trendlines