Fig. 4

Contamination estimates on simulated datasets. The datasets differ in the number of sequences (a), deamination rates (b) and sequencing error/divergence rates (c). Each point corresponds to a set of simulated sequences with varying proportions of contamination (x-axis) from 0 to 95% in steps of 5%. The errors (y-axis) correspond to the difference between the estimated and true contamination rates. The bars correspond to the 95% CI