HomoLigAlign
HomoLigAlign
refer to section "Pose Prediction Method"
Ligand tautomeric/ionization states and 3D coordinates were generated with LigPred from Schrodinger suite. Homology models with Modeller. Embeded ligand conformers with RDKit. MD simulations with AmberTools and OpenMM.
For all 20 ligands ligand a 100 ps MD simulations was conducted to equilibrate each system.
HomoLigAlign algorithm (still unpublished) combines homology modeling, ligand alignment and MD to generate protein-ligand poses for scoring. The algorithm starts by quering PDB for homologous proteins to the target receptor co-crystalized with ligands that are similar to those that we wish to predict their binding pose (screening compounds). Upon selecting the most appropriate structures for each screening compound, it creates a homology model of the target receptor with the original ligand. It then generates multiple conformers of the screening compound embeded on the original ligand by matching coordinates of the atoms in the maximum common substructure between those 2 molecules. Then based on some energy, shape and RMSD criteria, it selects the best embeded conformer and performs 100 ps of MD simulation in explicit solvent to equilibrate the system. The final ligand pose is the last frame of the 100 ps trajectory upon a 1000 step minimization.
Yes
Yes