Neurological diseases
Alzheimer's disease
Amyloid-β plaques and tau tangles that erode memory and cognition.
Parkinson's disease
α-synuclein aggregates that disrupt movement and balance.
Huntington's disease
An expanded huntingtin protein that progressively kills neurons.
Why the brain is especially vulnerable
Most cells in your body are replaceable — skin, gut lining, blood. Neurons aren't. The brain you're using to read this is largely the one you'll have for life, and the proteins inside it have to keep working for decades.
When a protein in a neuron misfolds, the damage doesn't get cleared by replacing the cell. It accumulates. Alzheimer's, Parkinson's, and Huntington's all share that pattern: misfolded proteins building up over years, breaking the machinery that keeps neurons alive.
When proteins can't fold
A protein has to assemble itself into a precise 3D shape to do its job. When folding fails, copies of the misfolded protein often stick to each other, forming clumps that grow into fibrils. Once the process starts, more copies pile on.
Those clumps are the plaques and tangles of Alzheimer's, the Lewy bodies of Parkinson's, and the toxic aggregates of Huntington's. They're not just byproducts of the disease — they actively kill the neurons around them.
When proteins flicker between shapes
Folding isn't the only thing that can go wrong. Many proteins work by switching between several folded shapes — opening and closing, binding and releasing. Disrupt that rhythm, and the protein still folds correctly but stops doing its job.
The NMDA receptor is one of the most important examples. It's a channel in the membrane of a neuron that opens and closes to let signals through, and it sits at the heart of how we form memories. When it opens too often, it contributes to epilepsy, Alzheimer's, Parkinson's, and stroke damage. When it opens too rarely, it's implicated in schizophrenia.
Using Folding@home, we found a new way that NMDA receptors control their own switching, and our lab partners confirmed it experimentally. Every donated CPU cycle is what makes long-timescale simulations like this possible.