Small molecule drug and biotech drug interaction prediction based on multi-modal representation learning

BMC Bioinformatics

Table 2 The performance of Multi-SBI with different feature combinations

Method	ACC	AUC	AUPR	F1	Pre	Rec
CNN	0.9336	0.9993	0.9794	0.8016	0.8111	0.8221
daylight/EMS	0.9106	0.9991	0.9652	0.8047	0.8683	0.7983
SPI/BPI	0.7736	0.9960	0.8770	0.4772	0.5741	0.4577
SSI/BBI	0.8211	0.9976	0.9190	0.5623	0.6503	0.5219
CNN + daylight/EMS	0.9427	0.9995	0.9807	0.8337	0.8569	0.8410
CNN + SPI/BPI	0.9353	0.9992	0.9705	0.8005	0.8302	0.8028
CNN + SSI/BBI	0.9381	0.9993	0.9745	0.8131	0.8450	0.8096
daylight/EMS + SPI/BPI	0.9423	0.9994	0.9817	0.8462	0.8862	0.8259
daylight/EMS + SSI/BBI	0.9413	0.9994	0.9803	0.8167	0.8645	0.8070
SPI/BPI + SSI/BBI	0.8809	0.9985	0.9524	0.6384	0.7016	0.6135
daylight/EMS + SPI/BPI + SSI/BBI	0.9410	0.9994	0.9810	0.8399	0.9003	0.8208
CNN + daylight/EMS + SPI/BPI	0.9492	0.9996	0.9860	0.8461	0.8627	0.8506
CNN + daylight/EMS + SSI/BBI	0.9490	0.9996	0.9859	0.8582	0.8741	0.8577
CNN + SPI/BPI + SSI/BBI	0.9404	0.9994	0.9791	0.8120	0.8467	0.8086
CNN + daylight/EMS + SPI/BPI + SSI/BBI	0.9676	0.9997	0.9892	0.8673	0.9039	0.8509

ISSN: 1471-2105