. 2021 Mar 29;17(3):e1008864. doi: 10.1371/journal.pcbi.1008864

Table 6. Classification of Abs (ADI-15734 and ADI-15878) recognizing two epitopes at the base of the EBOV-GP trimer.

We colored the fingerprints for the DNN models according to the charge coloring code.

No. DNN models	Training and Validation			Testing Results
No. DNN models	‹N_training›^a	‹M_validation›^b	‹A_val›^c	‹N_test›^d	‹N_correct›^e	‹A_test›^f	‹κ›^g
30	893 (SD 64)	269 (SD 34)	1.00 (SD 0.01)	478 (SD 76)	420 (SD 79)	0.88 (SD 0.10)	0.79 (SD 0.19)

^a N_training; number of fingerprints selected for training of the model.

^b M_validation; number of fingerprints selected for validation.

^c A_val; validation accuracy.

^d N_test; number of images in the testing sets.

^e N_correct; number of images in the testing sets predicted correctly.

^f A_test; accuracy as evaluated using the test sets.

^g κ; Cohen’s Kappa coefficient for predictions of the testing sets.