/drug_binding_site

Primary LanguageJupyter Notebook

Binding site ID pipeline:

  1. Take any protein, feed through CHARMM-GUI to generate an input file. The final step (step3 for solvated, step5 for membranated) stepX_assembly.pdb is required for 3) below.

  2. Run drug_parameterization.ipynb: Openforcefield to generate ligand parameterization. Just input the isomeric smiles code and RDKit + openforcefield will generate correct pdbfile and AMBER forcefield-compatible parameters

  3. Run system_setup.ipynb: System setup. This combines the drug system with the membranated protein system, and gradually releases constraints on the protein over ~25 ns. Drug molecules have a pairwise repulsive bias that drops off exponentially with distance to prevent aggregation. For solvated proteins (i.e. KRAS and T4 Lysozyme) the drugs are ranged all along the periodic boundary conditions while ensuring no clashes. For membrane systems (i.e. GlpG), the zrange is confined to the bilayer because the ligand is hydrophobic and we are not interested in measuring bilayer partitioning.

  4. Determine the binding site. Suggest removing the protein restraint completely, or at least replacing with an alpha-carbon only restraint. Use any combination of simulated tempering, generalised-REST, metadynamics, etc. determine_weights.ipynb will equilibrate weights for either simulated tempering or serial g-REST

  5. Run analysis.ipynb to calculate relative binding site preferences based on histograms.