Table 3.

The performance of combinations of top-scoring features from each group of features.

Feature combination	Test dataset	Xanthomonas data
Amino acids frequencies + LME	0.81 (0.83)	0.70 (0.72)
LME + HMM	0.8 (0.8)	0.68 (0.7)
LME + Features from Hobbs et al. (2016)	0.83 (0.91)	0.72 (0.87)
Amino acids frequencies + LME + HMM	0.81 (0.81)	0.7 (0.7)
Amino acids frequencies + LME + Features from Hobbs et al. (2016)	0.8 (0.81)	0.7 (0.72)
HMM + LME + Features from Hobbs et al. (2016)	0.82 (0.85)	0.71 (0.73)
Amino acids frequencies + LME + HMM + Features from Hobbs et al. (2016)	0.82 (0.83)	0.72 (0.72)

The results were obtained on the test datasets by analyzing the 100 N-terminal amino acids. The scores are MCC and AUPRC (in parenthesis). In bold are the best scores for each of the test datasets.