To map the protein interaction network in the yeast Saccharomyces cerevisiae, for instance, researchers used sophisticated techniques to knock out (disable) pairs of genes, one pair at a time, creating doubly mutant cells. They then compared the fitness of each double mutant (based in part on the size of the cell colony it formed) to that predicted from the fitnesses of the two single mutants. The researchers reasoned that if the observed fitness matched the prediction, then the products of the two genes didn't interact with each other, but if the observed fitness was greater or less than predicted, then the gene products interacted in the cell. Computer software then mapped genes based on the similarity of their interactions; a network-like "functional map" of these genetic interactions is displayed in Figure 21.5. To process the vast number of protein-protein interactions generated by this experiment and integrate them into the completed map required powerful computers, mathematical tools, and newly developed software.
