Fig. 4.

Performance comparison on CDS construction among inGAP-CDG and 11 other combined pipelines. a The length distribution of predicted CDSs by each method. b The ROC curve of each method. ROC, which captures the trade-offs between sensitivity and specificity, is determined by calculating the true-positive and false-positive rates. c The redundancy of predicted CDSs by each method. The redundancy is calculated by the number of aligned CDSs divided by the number of reference genes. d The evaluation of CDS fragmentation in each method. inGAP-CDG outperforms the other 11 pipelines, with an average of 1.18 fragments per gene