Skip to main content

Table 8 The performance of different network architectures on the SignalP Eukaryotes dataset

From: SigUNet: signal peptide recognition based on semantic segmentation

Architecture

MCC (%)

FPRTM (%)

Recall (%)

Precision (%)

F1 measure (%)

U-Net-1Da

84.1

6.8

87.3

88.5

87.9

SigUNet-maxb

88.6

5.0

91.4

91.3

91.3

SigUNet

90.2

4.0

92.1

93.0

92.5

  1. aA one-dimensional U-Net that has the same network configuration as Fig. 3, but the input and output layer are modified for protein sequences. bThe max pooling layers in Fig. 4 are replaced with average pooling layers