My lab works on proteins. A few of the people run "dry" experiments, meaning they enter equations and run simulations of molecules and how they interact with proteins. Needless to say, being as inept with computers as I am, I normally don't deal with that.
But recently I was giving a box of 32 compounds that I had to test for allosteric activity. According to the computers, they were all allosteric enhancers.
According to my experiments, they're not.
Admittedly, a hit rate of 3-4 out of 32 isn't half-bad. Random screening without using computers to narrow down the list of criteria for our allosteric compounds would have me pulling my hair out of pure stress--when you're testing that many compounds, you have to make sure that the labels on one tube matches the label on the other, or else...
The point of all this is that life doesn't compute well. We can plug data into our computers until we're blue in the face, and get all types of random, seemingly meaningless correlations (rainy weather, autism), or important, seemingly significant ones (cholesterol, heart attacks). But it takes getting your hands dirty--designing the surveys, running the statistics, cracking the math--before you might turn up something useful.
And all too often, you don't.
I've still got a few more tests to run, but at this point they're more for verification of what I already know (most of the compounds are not allosteric modulators) than to get new data.