SigUNet: signal peptide recognition based on semantic segmentation

BMC Bioinformatics

Table 1 Statistics of the datasets that are used in this study

Organism	Signal Peptides		Transmembrane		Cytosolic or Nuclear		Total
Organism	Train	Comp	Train	Comp	Train	Comp	Total
SignalP
Eukaryotes	1640	606	987	939	5133	1000	7760
Gram-positive	208	48	117	117	360	213	685
Gram-negative	423	104	523	523	912	260	1858
SPDS17
Eukaryotes	–	46	–	323	–	689	1058
Gram-positive	–	9	–	189	–	240	438
Gram-negative	–	23	–	89	–	99	211

The SignalP dataset is from the UniProtKB/Swiss-Prot in accordance with the identity list in Pertersen et al.’s study [12]; The SPDS17 dataset is from the UniProtKB/Swiss-Prot in accordance with the identity list in Savojardo et al.’s study [6].

ISSN: 1471-2105