From Viral Protein to Antiviral Lead: An AI-Assisted In Silico Design Journey

The COVID-19 pandemic has underscored the role of structural and computational biophysics in antiviral discovery. In this talk, I will present an in silico design workflow that connects viral protein characterization, molecular dynamics simulations, and AI-based modeling. We focus on two SARS-CoV-2 proteins – the multifunctional helicase NSP13 and the interferon antagonist ORF6 – as potential drug targets. I will discuss computationally identified and experimentally confirmed inhibitory motifs and introduce the concepts of protein aptamers and cyclotides as promising scaffolds for peptide-based antivirals. Using molecular dynamics simulations and a convolutional neural network (CNN) trained for interaction assessment, we have explored the feasibility of grafting viral peptide motifs onto cyclotide scaffolds. Although this approach is still under experimental validation, the results demonstrate how AI-enhanced in silico methods can transform structural insights into rational antiviral design strategies.