June 18, 2026 ~1 minute minutes read Marc

MD Simulations: Exploration vs. Justification?

Molecular dynamics simulations are often misused as definitive evidence in structural biology when they should be treated as hypothesis generators requiring exp…

Flipping a coin to decide dinner is harmless. Flipping a single 10 ns MD run to explain protein folding? That's a problem.

I love molecular dynamics. But I've noticed something unsettling: MD has become the coin flip of structural biology. Unsure about a mechanism? Throw a simulation at it. One trajectory. One window of time. One set of force field assumptions. Call it "ground truth."

Trajectories are proposals, not proof. They're hypothesis generators. The real power of simulation emerges when you run multiple conditions, aggregate observables, and crucially, when you design experiments to test what you observed. Not the other way around.

We need to stop treating a single (definitely much much longer than the mentioned provocative) 10 ns run as confirmation. It's a starting point. A conversation starter between computation and experiment. The moment you treat it as definitive, you've lost the scientific thread.

Let me ask directly: Are we using simulations to explore or to justify? Because there's a massive difference, and our papers should make that distinction clear.