The latest issue of Nature (08/16/2007) has an interesting cover headline: "Form Finds Function" introduces an article by Johannes Hermann called "Structure-based activity prediction for an enzyme of unknown function" that argues, in effect, that the function of a certain enzyme can be predicted on the basis of its underlying structure. This is an interesting finding to Aristotelians, who think that there are such things as essences in nature that do, in fact, determine the functions of things.
It also reminds me of a rather interesting factoid that I only learned about because my wife is a crime-novel buff. It seems that there is a certain compound that is often used in cough-syrups called dextromethorphan. This compound is identical to another, called levomethorphan, in every way except one: dextromethorphan twists to the right (hence the "dextro" part of its name) but levomethorphan twists to the left. Apart from the direction of their twisting they are indiscernible. Oh, actually, there is one other difference between them: dextromethorphan relieves your cough, but levomethorphan can kill you if you take too much of it. One is antitussive at low doses and dissociative at high doses, the other is an opiate that relieves pain at low doses and kills at high doses. That's why these compounds figure in crime novels: because they are chemically identical they cannot be distinguished when they are in your body doing things to you, so somebody who had access the the left-leaning molecule could kill you without being caught. I told you leftists were dangerous.
To make a long story short, it seems that we have some grounds here for thinking that strictly reductive, materialist accounts of systems are insufficiently explanatory of mechanism: form is also a necessary component of a full explanation.