Table 3. Performance of CNN, LSTM and RF models for binary classification when random sequences from human genome are used as negative class data with test data.

Methods	Human
	Acc	Se	Sp	MCC
	2-mer
CNN	0.9	0.85	0.96	0.84
LSTM	0.56	0.09	0.96	0.01
RF	0.75	0.77	0.72	0.49
	4-mer
CNN	0.91	0.88	0.99	0.85
LSTM	0.89	0.82	0.96	0.79
RF	0.79	0.77	0.81	0.58
	8-mer
CNN	0.91	0.85	0.98	0.83
LSTM	0.91	0.84	0.99	0.84
RF	0.82	0.77	0.86	0.64