Exploring Markov Processes Variational Approach for Dynamics Prediction of Antiviral Peptides

This work explores the application of the Variational Approach for Markov Processes (VAMP) in predicting the conformational dynamics of the antiviral peptide DENV-2 (CGYGLC) in an aqueous solution. A 10 ns molecular dynamics simulation is performed using AMBER, analyzing dihedral angles as the primary structural descriptors. Using VAMP, the temporal evolution of these angles is modeled through an approximation of the Koopman operator. Canonical correlation matrices are constructed to estimate the best representation of the system's slow dynamics. Finally, the dihedral angle values are predicted using a Singular Value Decomposition (SVD) applied to the evolution of molecular coordinates.