I'm going to write a sort-of decent explanation of what I'm doing this summer, so I have an easier time explaining it to people later.
I have an unpaid internship at Caltech doing computational chemistry, modeling 3-d shapes of proteins. Knowing the shape of proteins is important to understand how they react to chemicals, electrical impulses, body tissue, plasma cannons, &c. Proteins are comprised of chains of amino acids. The chain can be less than ten or more than ten thousand. However, they don't stay as one long chain. They fold, according to chemical structures, temperature, acidity, and other constraints until the structure looks like any number of amorphous blobs smooshed together.
Now, proteins are tiny little things, so the shapes can only be really determined by inputting the chain and a ridiculous number of other variables into a computer simulation and then waiting weeks for it to fold. However! Many proteins have similar structures and functions, and follow the "you've seen one you've seen them all" rule. By using comparisons, it is possible to take the already-formed protein, replace the dissimilar amino acids with the ones from the mystery-protein's sequence, and then fold that. This saves a lot of time and makes it far more efficient.
Well, I think it's cool.
