This was a graded course lab, not a published study. The design below is a proposed experiment, not one that was actually run — the docking results themselves are real output from an in-class exercise.
Molecular Docking
Modeling How a Drug Binds a Moving Target: Docking Tamiflu Against Shifting Flu Neuraminidase Strains
Context
What the system looked like before.
As I learned how to model 2D neighboring data points — in time, or along a chromosome — a course on analytical techniques in human disease began translating to me how a drug molecule physically interacts with a protein's binding pocket, based on atomic distances, energy states, and molecular constraints. Computation was a tool to simulate the fundamental biophysics of pharmacology — predicting exactly how a mutation in a virus might render a life-saving drug ineffective.
In plain terms: the course used molecular docking — computationally testing how well a small drug molecule fits into a protein's binding pocket — to study neuraminidase, an enzyme on the flu virus that lets new viral particles escape an infected cell. Drugs like Tamiflu (oseltamivir) work by jamming that pocket; different flu strains carry mutations there that can change how well the same drug binds.
Question
Given a flu mutation (S247N) that resists two neuraminidase-inhibiting drugs but not a third, how would you design a docking study to confirm whether peramivir really does bind better — without that conclusion depending on how you set up the simulation?
Approach & rationale
To better uncover the mechanisms of peramivir binding to S247N, I would design an experiment that compared the binding of peramivir, oseltamivir, and zanamivir to S247N using three different box dimensions for each ligand — one small, sized to the proposed binding site based on active-site residue separation; one medium; one large. I'd use these to determine the binding affinity of each ligand, compare results at each volume between ligands to see if peramivir is truly a better binder, and compare the effect of varying volume for each ligand to determine what dimensions serve as a better basis for the theoretical binding site.
Methodology & iteration
Including the paths that did not hold.
The guided lab
On search-box size
In short
Top binding-affinity hits (kcal/mol, more negative = more stable): N1 + oseltamivir −6.756, N2 + oseltamivir −5.731, N1 + laninamivir −7.107, N2 + laninamivir −5.594..
Results
Numbers first, not buried in prose.
Top binding-affinity hits (kcal/mol, more negative = more stable): N1 + oseltamivir −6.756, N2 + oseltamivir −5.731, N1 + laninamivir −7.107, N2 + laninamivir −5.594.
Based on the top affinity table, N1 is most affected by binding to oseltamivir; N2 is least affected — the more negative kcal/mol value for N1 indicates more free energy released on binding, a lower-energy, more stable state than the N2–oseltamivir complex. Laninamivir is the more effective ligand against N1 compared to oseltamivir, given its lower kcal/mol value. For N2, oseltamivir is the more effective ligand — it binds N2 with a slightly lower affinity than laninamivir does.
Significance
Where this went.
Coursework — not presented or published beyond the course.
Reflection
Computation became a tool to simulate the fundamental biophysics of pharmacology, letting me model exactly how a mutation in a virus might render a drug ineffective — a topographical map, not unlike the ones in genomics lab, of how structural relationships at every scale determine a functional outcome.