Table 9. Performance comparison of BiComp encoding, against LZMA and SW encodings, for drug-target binding affinity prediction, for Davis and Kiba datasets, using feature ablation experiments.

Experiments	Davis				Kiba
	CI(std)	MSE	${\mathbf{r}}_{\mathbf{m}}^{\mathbf{2}}$(std)	AUPR(std)	CI(std)	MSE	${\mathbf{r}}_{\mathbf{m}}^{\mathbf{2}}$(std)	AUPR(std)
Feature ablation SW	0.869 (0.001)	0.284	0.541 (0.014)	0.623 (0.005)	0.886 (0.001)	0.168	0.713 (0.016)	0.812 (0.004)
Feature ablation LZMA	0.885 (0.001)	0.279	0.597 (0.015)	0.681 (0.003)	0.877 (0.002)	0.177	0.693 (0.014)	0.803 (0.005)
BiComp-DTA	0.904 (0.001)	0.237	0.696 (0.012)	0.753 (0.006)	0.891(<0.001)	0.167	0.757 (0.012)	0.834 (0.003)

Values of CI, MSE, $r_m^2$ and AUPR are provided for BiComp compared to those of LZMA and SW encodings for Davis and Kiba datasets.