Fig. 4: Source effect evaluation of negative TCRs on retrained models.
From: Assessment of computational methods in predicting TCR–epitope binding recognition
a,b, Performance of PS- and HS-based retrained models evaluated on AS-based seen-epitope (a, n = 24) and unseen-epitope (b, n = 21) test data using CDR3β-only features. c,d, Performance of AS-based retrained models on PS- and HS-based seen-epitope (c, n = 24) and unseen-epitope (d, n = 21) test data with CDR3β-only features. Lines connect the same models across evaluation settings. All P values were from two-sided Wilcoxon signed-rank tests with Benjamini–Hochberg correction.
