Skip to main content

Table 3 The performance on the Gram-positive datasets

From: SigUNet: signal peptide recognition based on semantic segmentation

Method MCC (%) FPRTM (%) Precision (%) Recall (%) F1 measure (%)
The SignalP dataset
 Phobius 67.7 20.5 60.0 87.5 71.2
 PrediSi 40.9 54.7 35.0 75.0 47.7
 SignalP3.0-HMM 55.8 43.6 44.3 89.6 59.3
 SignalP3.0-NN 47.2 56.4 34.9 91.7 50.6
 PolyPhobius 71.1 16.2 66.1 85.4 74.5
 Philius 69.6 15.4 64.1 85.4 73.2
 SPOCTOPUS 73.9 15.4 67.2 89.6 76.8
 SignalP 4.0 85.1 2.6
 TOPCONS2 81.6 6.8 80.8 87.5 84.0
 DeepSig 73.9 6.8 81.4 72.9 76.9
 SigUNet 76.1 5.1 85.4 72.9 78.7
The SPDS17 dataset
 Phobius 35.0 13.6 17.9 77.8 29.2
 PrediSi 14.3 64.0 5.0 77.8 9.5
 SignalP3.0-HMM 27.3 27.0 11.9 77.8 20.6
 SignalP3.0-NN 16.1 45.5 5.7 77.8 10.7
 PolyPhobius 34.5 13.2 17.5 77.8 28.6
 Philius 30.3 79.0 16.2 66.7 26.1
 SPOCTOPUS 30.3 13.8 16.2 66.7 26.1
 SignalP 4.0 50.3 0.0 40.0 66.7 50.0
 TOPCONS2 38.1 4.2 24.0 66.7 35.3
 DeepSig 54.5 0.1 46.2 66.7 54.4
 SigUNet 40.9 2.1 40.0 44.4 42.1
  1. The best performance is highlighted in bold