Table 2.

Binary classification performance of functional phenotypic datasets using fine-turned models

	Model	Accuracy	F1	Loss	Precision	Recall	Auroc_macro	Pr_auc	MCC	Time(h)
Dataset_HERV	A	0.9354 ± 0.0025	0.9386 ± 0.0025	0.1748 ± 0.0026	0.9544 ± 0.0027	0.9234 ± 0.0051	0.9851 ± 0.0007	0.9883 ± 0.0006	0.8710 ± 0.0048	13.2393 ± 0.3802
	B	0.9363 ± 0.0010	0.9395 ± 0.0010	0.1554 ± 0.0012	0.9566 ± 0.0035	0.9230 ± 0.0040	0.9859 ± 0.0003	0.9888 ± 0.0003	0.8731 ± 0.0019	17.7381 ± 1.2064
Dataset_Immuno	A	0.8122 ± 0.0002	0.8051 ± 0.0037	0.3211 ± 0.0013	0.8414 ± 0.0132	0.7729 ± 0.0182	0.9238 ± 0.0007	0.9298 ± 0.0007	0.6274 ± 0.0017	0.7026 ± 0.0165
	B	0.8129 ± 0.0008	0.8128 ± 0.0039	0.3198 ± 0.0021	0.8182 ± 0.0167	0.8095 ± 0.0244	0.9238 ± 0.0013	0.9300 ± 0.0011	0.6274 ± 0.0018	1.9810 ± 0.0044
Dataset_Regulatory	A	0.7438 ± 0.0003	0.7764 ± 0.0028	0.5016 ± 0.0013	0.7540 ± 0.0060	0.8005 ± 0.0127	0.8270 ± 0.0005	0.8602 ± 0.0004	0.4787 ± 0.0009	18.4481 ± 0.0137
	B	0.7614 ± 0.0012	0.7911 ± 0.0020	0.4744 ± 0.0015	0.7704 ± 0.0061	0.8134 ± 0.0104	0.8462 ± 0.0014	0.8763 ± 0.0012	0.5147 ± 0.0027	27.2533 ± 1.6940
Dataset_Diseases_GWAS	A	0.8380 ± 0.0007	0.8537 ± 0.0021	0.3023 ± 0.0023	0.8642 ± 0.0095	0.8442 ± 0.0130	0.9368 ± 0.0008	0.9541 ± 0.0006	0.6731 ± 0.0022	7.2913 ± 0.0123
	B	0.8443 ± 0.0010	0.8609 ± 0.0014	0.2976 ± 0.0019	0.8622 ± 0.0048	0.8597 ± 0.0070	0.9393 ± 0.0008	0.9555 ± 0.0006	0.6843 ± 0.0018	21.3534 ± 0.0286
Dataset_Highly_Specifically_Gene	A	0.6346 ± 0.0044	0.6081 ± 0.0180	0.6075 ± 0.0029	0.6688 ± 0.0081	0.5604 ± 0.0352	0.7076 ± 0.0034	0.7372 ± 0.0043	0.2758 ± 0.0054	0.2051 ± 0.0003
	B	0.6237 ± 0.0056	0.5861 ± 0.0147	0.6349 ± 0.0040	0.6668 ± 0.0189	0.5264 ± 0.0332	0.6894 ± 0.0011	0.7261 ± 0.0011	0.2572 ± 0.0144	0.1858 ± 0.0098
Dataset_Random	A	0.5205 ± 0.0031	0.6503 ± 0.0118	0.6953 ± 0.0008	0.5299 ± 0.0006	0.8436 ± 0.0384	0.4983 ± 0.0021	0.5296 ± 0.0006	-0.0003 ± 0.0026	0.7690 ± 0.0038
	B	0.5134 ± 0.0059	0.5895 ± 0.0242	0.7010 ± 0.0014	0.5327 ± 0.0019	0.6656 ± 0.0606	0.4997 ± 0.0012	0.5290 ± 0.0014	0.0085 ± 0.0067	2.4468 ± 0.0038

Note The A and B represent the DNA_bert_6 model and the human_gpt2-v1 model, respectively