Computational Drug Design: HIV Protease Docking

Research question

How can molecular docking be used to model the binding of lopinavir to HIV-1 protease, and how do grid settings affect predicted binding conformation and interaction analysis?

Background

This was a Grade 11 research project at Shanghai Foreign Language School Affiliated to SISU, Minhang Foreign Language School. It studied HIV-1 protease as a drug target and lopinavir as an HIV protease inhibitor. The project later received First Prize in the 2021 Shanghai Youth Science and Innovation Workstation Competition.

Model and tools

The project used Schrödinger Maestro for ligand preparation, receptor preparation, docking-grid generation, and Glide docking. The receptor was HIV-1 protease structure 1MIU, and the ligand was lopinavir.

Methods

The workflow prepared lopinavir in 2D and 3D, optimized the HIV-1 protease receptor structure, generated docking grids at 15, 18, and 20 Å, docked lopinavir under each condition, ranked conformations by Glide GScore, and interpreted hydrogen-bond, hydrophobic, and electrostatic interactions.

Implementation

The public artifact is a cleaned English portfolio version of the original Chinese report and presentation. I converted the workflow into a LaTeX report and a Beamer deck, keeping the computational claims bounded to the original docking output and avoiding unsupported biomedical conclusions.

Results

The strongest reported score was obtained under the 18 Å grid setting: conformation LOP_18_1 with Glide GScore -11.898. The 15 Å and 20 Å best poses were LOP_15_1 with -10.758 and LOP_20_1 with -10.192.

Validation

The project compared multiple grid sizes and multiple top-ranked poses rather than relying on a single docking result. The interpretation also considered interaction mechanisms, not only docking scores.

Limitations

This is a high-school computational project, not a validated biomedical or pharmacological result. Docking scores are sensitive to receptor preparation, protonation state, grid selection, force-field settings, and software assumptions. No molecular dynamics, free-energy calculation, or experimental validation was included.

What I learned

The project introduced me to computational modelling, molecular visualization, scientific literature reading, and the distinction between producing software output and making a validated scientific claim.

Future work

Future work could compare more HIV protease inhibitors, document the exact input files and parameters, add molecular-dynamics simulations, and build a reproducible repository for docking workflows.