Figure 1 | Scientific Reports

Figure 1

From: Optimizing variant-specific therapeutic SARS-CoV-2 decoys using deep-learning-guided molecular dynamics simulations

Figure 1

Multi-level strategy to identify high-affinity human ACE2 (hACE2) variants for neutralization of diverse SARS-CoV-2 variants. In brief, homology models (HM) of varying SARS-CoV-2 spike receptor binding ___domain (RBD) structures in complex with hACE2 variant structures served as input for molecular dynamics (MD) simulations analyzed via an empirical scoring function (ESF) closely related to the linear interaction energy (LIE) model. Gibbs free energy predictions (ΔG) were combined with Catalophore Halos to train an artificial neural network (ANN). This enables the model to predict ΔG values for hACE2- and SARS-CoV-2 variants based on their Halos. ANN ΔG predictions were validated with ESF ΔG values. The SARS-CoV-2 neutralizing potential of promising hACE2 variants was evaluated in ELISA-based competitive inhibition assays followed by live SARS-CoV-2-based neutralization assays.

Back to article page