Fig. 2: Computational metrics of sequences experimentally tested in round 2.
From: Computational scoring and experimental evaluation of enzymes generated by neural networks
a, AUC-ROC scores of activity versus metrics. The average of CuSOD and MDH. b, Spearman correlations between metrics. The average of CuSOD and MDH. c, Empirical cumulative distribution functions of selected metrics for active (solid lines) and inactive (dashed lines) sequences. Curves represent the pooled results of all generative models and both enzyme families.
