Fig. 6

Comparison of the SV calling performance using the recommended datasets and 26 other datasets. Relationship between the recall and precision of variant calling tools on 29 datasets a without and b with genotyping. The Coverage, Length and Error legends indicate that the datasets vary in sequencing coverage, average read length and sequencing error rate, respectively. Recommendation-1 represents datasets with 10% and 7.5% error rates, 20× sequencing coverage and 20 kbp read length, and Recommendation-2 represents the dataset with a 1% error rate using the same coverage and read length. Datasets in the orange circular area have better performance