Effect of the allele sampling bias on allele frequency estimates. Accuracy and precision of the allele frequency estimates under biased sampling of alleles from heterozygous individuals, with accurate or inaccurate estimates of the allele sampling bias. The titles above the plots show the real values of λ. A red line is traced where the estimated and current allele frequencies are the same. Each combination of real and estimated λ values is simulated 1000 times, using random allele frequencies. All allele frequency estimates are run here with 1000 individuals sequenced at 4 × coverage, in order to reduce the dispersion of the estimates and make the bias caused by the misspecification of λ more visible.