Fig. 2: Accuracy of software performance on simulated datasets of Nanopore R103, PacBio SEQUEL CLR, and CCS.

Precision (A) and sensitivity (B) of the tested methods on simulated data from Nanopore R103, PacBio SEQUEL CLR, and CCS across varying read depths (10X, 25X, 40X, 55X, 70X, with three replicates for each sequencing platform, totaling n = 45). Precision (C) and sensitivity (D) of the performance of tested methods obtained on simulated data of Nanopore R103, PacBio SEQUEL CLR, and CCS across different transcriptome complexity indices (1, 3, 5, 7, 9; values in parentheses denote the actual mean number of isoforms per gene simulated), with three replicates per sequencing platform (n = 45 in total). Precision (E) and sensitivity (F) of the performance of tested methods obtained on simulated data of Nanopore R103, PacBio SEQUEL CLR, and CCS with different read completeness (0.0_0.0: 100% complete, 0.1_0.1: 10% truncated from both ends; 0.2_0.2: 20% truncated from both ends; 0.2_0.0: 20% truncated from 5’ end; 0.4_0.0: 40% truncated from 5’ end; 0.0_0.2: 20% truncated from 3’ end; 0.0_0.4: 40% truncated from 3’ end, three replicates for each sequencing platform, n = 63 in total). Precision (G) and sensitivity (H) of the performance of tested methods on simulated data of Nanopore R103, PacBio SEQUEL CLR, and CCS with different read accuracy (0.8, 0.85, 0.9, 0.95, 1, three replicates for each sequencing platform, n = 30 in total). Precision (I) and sensitivity (J) of the performance of tested methods on simulated data of Nanopore R103, PacBio SEQUEL CLR, and CCS with different annotation completeness (20%, 40%, 60%, 80%, 100%, three replicates for each sequencing platform, n = 45 in total). N and P represent datasets generated from the Nanopore and PacBio platforms, respectively. All reported values are expressed as means, with Standard Deviation (SD) detailed in the Source Data file. Source data underlying (A–J) are provided as a Source Data file.