So, basically what the paper linked from this article describes is a method of taking a set of interacting regulator genes and determining the minimum set of ge

So, basically what the paper linked from this article describes is a method of taking a set of interacting regulator genes and determining the minimum set of genes which have to be perturbed (activated or repressed) to move the overall expression pattern from one stable state to another. Their method involved grouping the genes into small sub-sets which have a small number of stable states. Their example is two genes which repress each other acting as a toggle switch. They then connect those "switches" which influence each other, forming a larger graph. The final step is to find the minimal set of genes (one from each strongly connected sub-graph of the DAG they just created) which will influence the entire network. So once you know the major regulatory genes involved in different cell types this method allows you to pick a small groups of genes whose expression has to be modified to convert between the two. Of course this entirely ignores the problem of coming with these regulatory networks in the first place. Many genes and networks of genes have already been characterized in scientific literature, but I think it's safe to say that there are many more that haven't. The issue being that there is no good way for determining gene interaction without lengthy in vivo analysis of each individual gene. There are other difficulties as well. As the authors mention, some effects of differentiation aren't reversible with this technique, e.g. epigenetic modification. This method, seems like it will gradually become more useful as our base of knowledge increases. Just don't expect it to do anything magical at the moment. And contrary to what TFT implies it is far from computer-controlled cell reprogramming.