Prediction of polyreactive and nonspecific single-chain fragment variables through structural biochemical features and protein language-based descriptors

Table 3 Performance of ensemble models with the trained model combinations

Name	Method	Test AUC	Accuracy	Precision	Recall	F1-score
A36	Average-based Ensemble (AVG)	0.836	0.758	0.750	0.703	0.726
A10		0.839	0.764	0.757	0.710	0.733
A5		0.839	0.760	0.755	0.701	0.727
A3		0.839	0.764	0.751	0.724	0.737
B9		0.838	0.754	0.751	0.688	0.718
B5		0.839	0.759	0.748	0.711	0.729
B3		0.840	0.765	0.755	0.717	0.735
A36	Linear Regression-based Ensemble (LR)	0.832	0.754	0.773	0.653	0.708
A10		0.828	0.748	0.745	0.680	0.711
A5		0.828	0.748	0.745	0.680	0.711
A3		0.831	0.759	0.751	0.706	0.728
B9		0.819	0.743	0.761	0.639	0.695
B5		0.830	0.758	0.750	0.705	0.727
B3		0.825	0.757	0.748	0.705	0.726
TAPE/GBM		0.824	0.745	0.741	0.679	0.712
F46/UniRep/GBM		0.834	0.759	0.751	0.706	0.728

ISSN: 1471-2105