FIGURE 2.

Examination of clustering algorithms. Parameter scans of eight clustering algorithms were performed using the Cao et al. (2023) training set. The performance of clustering was measured in terms of the number of antibodies in epitope-consistent multiple-occupancy clusters (y-axis). The maximum value of this metric achieved by a specific algorithm across all evaluated parameters when clustering the Cao et al. (2023) training set is shown. The ranges and optimal values of the evaluated parameter are shown in Table 1. The agglomerative clustering algorithm selected for SPACE2 is highlighted in blue.