Does anyone have more details on this? The article has no actual substance.
If I were to guess, it seems like it's doing something like a genetic algorithm -- try some things, see which ones do something interesting, try variations on those. But the article doesn't say.
From what I have seen after skimming through their site, they use multiple methods - logic programming solver in Prolog [1] and also some evolutionary algorithms [2].
If you are interested in details, they put online tons of supporting material for Science article (including Prolog sources) [3].
First, some terminology, w/ an attempt to keep it simple:
1) gene - sequence of DNA that can be used to make proteins
2) enzyme - protein that acts as a catalyst (hence, the relation to metabolism)
Without having yet read TFA (in /Science/), the meat of the article to me is this bit:
1) "They armed Adam with a model of yeast metabolism and a database of genes and proteins involved in metabolism in other species..."
Adam knows what DNA makes up the proteins that are enzymes for yeast metabolism (start here: http://www.ncbi.nlm.nih.gov/sites/gquery?term=yeast and you can probably find such information if you know what to look for)
Then, they selected some proteins involved in metabolism in other species (also probably able to find that on NCBI). I don't know how far away in the evolutionary family tree they might have gone with these "other species."
2) "Adam sought out gaps in the metabolism model, specifically orphan enzymes, which scientists think exist, but which haven't been linked to any parent genes. After selecting a desirable orphan, Adam scoured the database for similar enzymes in other organisms, along with the corresponding genes. Using this information, it hypothesized that similar genes in the yeast genome may code for the orphan enzyme."
Here it explained that it's not really a hypothesis. Well, it is in the sense that the hypothesis is 'X gene codes for the orphan enzyme.' Then it could test it having been programmed to do laboratory work (presumably) to do so.
I say "it's not really a hypothesis" because I would have considered the hypothesis of the broader experiment to be something along the lines of 'Proteins/Enzymes for whom we don't know which genes code may be coding for metabolism." Then the experiment is "find all such proteins and test them."
3) "Still chugging along on its own, it designed experiments to test its hypotheses, and performed them using a fully automated array of centrifuges, incubators, pipettes, and growth analyzers."
I don't know anything about the lab work, but before this part happens I gather:
a) Adam knows what genes it's looking at (it found them, after all)
b) It looks at those genes being turned into proteins (following the Central Dogma - http://en.wikipedia.org/wiki/Central_dogma_of_molecular_biol...)
c) Using thermodynamics or some aspect of chemistry, it can determine if those proteins play a role in metabolism.
Hope that helps. I'm still very much a novice at this myself.
I was very surprised first time I realized scientific research is a lot like... work. Not so much thinking as I imagined, and a lot of other stuff. So I'm not surprised to find out somebody took a domain where the work can be automated and did it.
If I were to guess, it seems like it's doing something like a genetic algorithm -- try some things, see which ones do something interesting, try variations on those. But the article doesn't say.