Reconstucting ancestral genes is a standard procedure. As I said, we did it as a class project in a freshman course in the chemical basis of evolution.
Very simple example: let's say you four living organisms have, for a protein fragment, amino acid sequences as follows
glsdgewqlv glsdgewqmv vlsegewqlv vltdaewhlv
Well, in third position in the chain, three organisms have an s and the last has a t, so the chances are the common ancestor was s, and we had a single mutation in the ancestral line of organism 4 to give a t. In the fourth position, three have a d and one has an e, so it's most likely the common ancestor had a d. In the fifth position, the common ancestor probably had a g, in the eighth position a q, and the ninth position an l. The first postion is evenly divided, so we can't tell, based on this data (based on a bigger data set, it's almost certainly a g). So the sequence for the common ancestor was likely:
[g/v]lsdgewqlv
Now for a short strand of protein and a few organisms, the probabilities are not definitive, but when you do it for hundreds of species, you can get the common ancestor sequence with very high probability of being correct, as well as the family tree. In the above example, for example, you can tell organisms 1 and 2 share a common ancestor, as do 3 and 4. In fact, the four organisms are the house mouse, the brown rat, the sperm whale, and the finback whale, and the protein is myoglobin.
What is new in the study is not the fact they deduced the common ancestor, it's that they made and expressed the gene. I've always thought this had incredible potential for studying the metabolism of organisms that became extinct millions of year ago.
Good point. I kind of implicitly went along with aMPU's (wildly false) suggestion that there is something new about inferring genetic sequences, or that this hadn't already been thoroughly examined. I should have noted that inferring ancestral sequences is (sometimes, depending on the specific method being used) an incidental step in other procedures such as inferring phylogenies or aligning sequences.
Correct me if I ere. I'm not terribly educated on the topic. Most of what I know I got from reading Kimura's book on "The Neutral Theory of Molecular Evolution". It discussed the various methods of inferring phylogenies and the like in a good bit of detail, but it's dated by now I'm sure, and years since I read it.