Skip to main content
. 2013 Apr 3;41(10):e108. doi: 10.1093/nar/gkt214

Table 6.

Performance of exon–exon junction detectors in junction detection and read mapping from using simulation data

Coverage Method Junctions
Junction reads
All reads
Time (min)
Rec (%) Acc (%) Rec (%) Acc (%) Rec (%) Acc (%)
30× Subjunc 92.4 98.3 90.5 98.0 93.2 96.1 2
MapSplice 93.1 97.3 86.3 95.4 95.8 88.9 15
TopHat 92.0 92.3 86.7 91.9 95.7 92.0 16
TopHat 2 92.4 93.4 87.6 93.5 96.4 89.0 15
70× Subjunc 93.2 98.0 90.7 98.0 93.3 96.1 3
MapSplice 94.0 97.0 86.3 95.4 95.8 88.8 17
TopHat 93.0 91.7 87.2 91.9 95.9 92.0 26
TopHat 2 93.5 93.0 88.1 93.4 96.6 88.9 24
100× Subjunc 93.3 98.0 90.8 98.0 93.3 96.2 5
MapSplice 94.3 96.9 86.3 95.5 95.9 89.0 18
TopHat 93.0 91.0 87.2 91.9 95.9 92.0 32
TopHat 2 93.7 93.0 88.1 93.6 96.6 89.0 30

Three simulation data sets were used, which have different sequencing coverages (30×, 70× and 100×). Column ‘Junction’ gives the recall rate and accuracy for the detection of exon–exon junctions by each method. Recall rate is the percentage of correctly reported junctions in all junctions generated in the simulation data, and accuracy is the percentage of correctly reported junctions in all reported junctions. Column ‘Junction reads’ (or ‘All reads’) gives the recall rate and accuracy of mapping of junction reads (or any reads). Recall rate is the percentage of correctly reported junction reads (or any reads) in all generated junction reads (or all reads) in the simulation data, and accuracy is the percentage of correctly mapped junction reads (or any reads) in all reported junction reads (or any reads). Results are given for Subjunc with default settings.