Table 4 Performance comparison of feature representation methods using their respective optimized CNN architectures for Khib site prediction in the human dataset. Boldface values indicate the best performance for each metric.

Feature representation method	10-fold cross-validation set				Independent test set
Feature representation method	ACC	F1	MCC	AUC	ACC	F1	MCC	AUC
ESM	0.724	0.725	0.448	0.793	0.718	0.737	0.439	0.784
One hot	0.745	0.756	0.492	0.806	0.741	0.763	0.487	0.811
CTD	0.741	0.774	0.505	0.806	0.715	0.752	0.445	0.787
PSSM	0.780	0.788	0.561	0.859	0.791	0.809	0.590	0.866
AAP	0.803	0.808	0.608	0.885	0.815	0.822	0.631	0.899
BLOSUM	0.818	0.825	0.640	0.902	0.823	0.837	0.653	0.913

Quick links

Search