Fig. 3: ELISA binding assay results for CR3022 and anti-5T4 UdAb mutants designed with a GearBind-based pipeline.
From: Pretrainable geometric graph neural network for antibody affinity maturation
On the left panels (a, c, e, g, i), the concentration-response curves alongside with EC50 values evaluated from ELISA assays are displayed, with the center denoting the mean absorbance and error bars indicating the standard deviation from three technical duplicates. On the right panels (b, d, f, h, j), the fitted EC50 values, their 95% confidence intervals and the fold changes in binding calculated as \({{{{\rm{EC}}}}}_{50}^{({{{\rm{wt}}}})}/{{{{\rm{EC}}}}}_{50}^{({{{\rm{mt}}}})}\) are displayed. Tested systems include the first-round CR3022 designs binding to Delta RBD (a, b); the second-round CR3022 designs binding to Delta RBD (c, d), Omicron S protein (e, f) and SARS-RBD (g, h); and single-point mutants of anti-5T4 UdAb (i, j) binding to 5T4.
