Skip to main content

Table 8 Performance comparison of DFFNDDS and competitive methods on the DrugcombDB dataset under leave-one-drugpairs-out splits

From: DFFNDDS: prediction of synergistic drug combinations with dual feature fusion networks

Method

ACC

BACC

Prec

Rec

F1

ROC AUC

MCC

Kappa

AP

DFFNDDS

0.739(0.002)

0.718(0.002)

0.759(0.003)

0.825(0.009)

0.791(0.003)

0.816(0.001)

0.447(0.003)

0.445(0.003)

0.864(0.002)

DeepDDS

0.822(0.004)

0.749(0.008)

0.756 (0.020)

0.575(0.017)

0.653(0.014)

0.856(0.007)

0.545(0.015)

0.536(0.014)

0.756(0.014)

DeepSynergy

0.795(0.006)

0.705(0.010)

0.730(0.008)

0.486(0.001)

0.583(0.006)

0.809(0.007)

0.471(0.016)

0.455(0.008)

0.693(0.001)

MRGNN

0.754(0.007)

0.670(0.009)

0.624 (0.017)

0.459(0.028)

0.528(0.019)

0.755(0.009)

0.376(0.016)

0.367(0.017)

0.606(0.017)

GCNBMP

0.652 (0.177)

0.576(0.093)

0.305(0.274)

0.315(0.260)

0.606(0.130)

0.161(0.198)

0.151(0.076)

0.160(0.196)

0.417(0.154)

EPGCNDS

0.750(0.003)

0.611(0.005)

0.704(0.004)

0.270(0.004)

0.389(0.007)

0.732(0.006)

0.319(0.004)

0.269(0.005)

0.573(0.013)

MatchMaker

0.805(0.003)

0.717(0.008)

0.738 (0.014)

0.508(0.018)

0.602(0.016)

0.822(0.007)

0.493(0.014)

0.479 (0.015)

0.708(0.017)

  1. The best performing model for each dataset and metric is highlighted in bold