Table 3.

TOUCAN best performances for the completely balanced (50% positive, 50% negative) CV models on A. niger test sets generated with a 10 000-amino acid sliding window

Training	Sliding			Post-
set	window	Overlap	Classifier	process	P	R	F-m
50%–50%	10 000	50%	svc	merge3	0.941	0.941	0.941
50%–50%	10 000	30%	randomf	merge3	1	0.965	0.982