The effect of sample size and sequencing depth. Boxplot of our estimated values minus the known value. The green/orange boxes indicate our full maximum likelihood method, whereas the other boxes are the genotype calling methods. We have generated 10 scenarios with and without selection, therefore each box represents different scenarios each with 100 data points estimated on the basis of the 100 × 1 Mb datasets. For these analysis we used a p-value of 10-6. Notice that the full ML method is mostly unbiased, but the variance is affected by the error rate and sequencing depth as expected. The genotype calling shows large biases that depend on sequencing depth, error rate and whether or not the region is under selection or not.