Why computing matters to medicine

A lot of modern medical research is, underneath, a computing problem. Working out the shape of a protein, sifting through genetic data, modelling how a molecule behaves, simulating a process too small or too fast to watch directly: all of it takes a great deal of computing, applied patiently over a long time. The science is done by researchers. What it increasingly needs alongside them is capacity, and capacity is something we can give.

What we actually do

We build computers and servers, so contributing computing is the natural thing for us to give. We keep a proactive research cluster of our own, always on, that takes its share of large distributed research efforts and returns the results to the shared pool. It is a quiet, steady thing, done alongside universities, volunteers and other companies who do the same. When a rare result shows up a little sooner because there were more machines on it, the pooled effort we are part of is some of the reason.

What it is

The cluster is a pool of the same machines we build, set aside and kept running for science. It does not wait around for spare moments. It is a proactive, always on contribution: it takes its share of large distributed research efforts, works through the pieces handed to it, and returns the results to the shared pool. Because it is our own hardware, we can keep it steady and predictable, which is exactly what long running research benefits from.

How large research efforts share the work: it is split into small pieces, run across many machines, and pooled back together. Our cluster is one of those machines.

How the work reaches us

Big research efforts do not need one enormous computer so much as a lot of computing applied patiently. The work is broken into small pieces, handed out to machines all over the world, and the answers are pooled. Our cluster takes a steady share of those pieces and sends its results back, one contributor among thousands.

Where our part fits

We are not doing all of this. Most of it is done by researchers with their own tools and their own expertise, and that is exactly as it should be. What ties these fields together is that they are hungry for computing, and computing is the thing we can add.

Our contribution is general purpose capacity, given to the efforts that pool it together. The clearest example, and the one we have written up in full, is protein folding, where the computing that reveals how a protein moves is some of the hardest to come by.

Shape is function

A protein folds into a three dimensional shape, and that shape is the reason it works. Its pockets and surfaces are what let it grab a molecule or lock onto a partner. Because the job depends on the shape, a small change can change the job, which is why a single genetic change can ripple outward into a protein that works differently, or not at all.

When a protein folds wrong and clumps, the result is tied to some of the hardest diseases in medicine, from Alzheimer's and Parkinson's to cystic fibrosis.

A still picture is not enough

Computers can now predict a protein's folded shape well. But a real protein is not frozen. It flexes and shifts, and a place where a drug could bind is sometimes hidden until the protein moves a certain way. Simulating that motion is where those hidden pockets show up, and it is the part that takes so much computing, which is where a research cluster like ours helps.

A predicted structure is one still picture. Simulation shows the motion, which is where hidden drug pockets appear.

How to read them

Each paper is written for a general reader, not only specialists. They are free, they cite their sources, and they are honest about how far the science has actually reached. If you only read one, start with the protein folding paper, which is the fullest account of the research we contribute to.