Incorporating significant amino acid pairs to identify O-linked glycosylation sites on transmembrane proteins and non-transmembrane proteins

Table 6 The five-fold cross-validation performance of O-linked glycosylation sites on transmembrane (TM) proteins and non-transmembrane (non-TM) proteins that are extracted from release 15.0 of UniProt.

Training features	Amino acid (BLOSUM62)		Amino acid + SAAPs		Amino acid + SAAPs + Membrane topology*
	TM proteins	non-TM proteins	TM proteins	non-TM proteins	TM proteins
True Positive	132	361	132	384	132
False Positive	462	2124	365	1933	317
True Negative	1988	12110	2085	12301	2133
False Negative	70	276	70	253	70
Sensitivity	65.3%	56.7%	65.3%	60.3%	65.3%
Specificity	81.1%	85.1%	85.1%	86.4%	87.1%
Accuracy	79.9%	83.9%	83.6%	85.3%	85.4%
Balanced Accuracy	73.2%	70.9%	75.2%	73.4%	76.2%
MCC	0.30	0.23	0.34	0.26	0.37

*This process considers the structural topology of transmembrane in the prediction of O-linked glycosylation sites on transmembrane proteins.

ISSN: 1471-2105