The subject of animal testing has been historically a bit controversial in the view of animal rights activists. Work with animals in academic and private research institutions is highly regulated. Whenever applying for funding for a study using animals as research subjects, researchers have to prove that the numbers proposed are the minimum they need to get a statistically valuable result, and that they will comply with all the approved protocols for care and management of the animals used.
Let's take as an example a mouse model that can be infected with a virus or bacteria that causes disease in a manner that can be monitored (a growth of a tumor, invasion of a tissue that can be dissected to evaluate the degree of infection, visible symptoms). This is the mouse model of the corresponding human disease
OK, so let's say a lot of research in a laboratory that specializes in studying the particular virus or bacteria of interest has resulted in a candidate gene mutation (if you go to the genetics and mutations section of my home page you can get more information on how this works). This research we are talking about here, let me emphasize, sometimes occurs in an academic setting where graduate students and postdoctoral researchers spend many days (and nights!) running experiments (sometimes again and again....) to learn more about possible genes/proteins to target in the infectious agent. Nowadays, compared to when I was a graduate student in the '90s, there is much more collaboration and communication between academia and private industry (pharmaceuticals) and biotech companies to combine efforts towards this goal. The private sector invests more money and resources in directly finding tools useful in prevention/cure whereas the academic laboratories might be researching at a more "basic" level. However, often the basic research which takes a long time and effort (and is usually not as well funded!) is what provides the basis for a candidate drug that eventually makes it to the market.
As I was saying before the disgression above, let's picture a scenario in which research has led to a candidate target for a drug and let's call this candidate protein Y. Protein Y is the product of the "expression" of the gene y. As I explained in the homepage, the beauty of genetics is that we can ask what the role of protein Y is by "mutating" gene y to result in the absence or a very diminished amount of its product protein Y. We can then use the mouse model to test the hypothesis that this gene/protein is indeed a good target. We infect the mouse with the mutant and ask whether this defective virus or bacteria can still produce the "phenotype" associated with the infection (in the figures, this phenotype is represented by the orange circle which could be a tumor). In the actual experiment, as with all serious ones when we evaluate any of these questions, there rare 2 groups of mice: one of them will be infected with the normal ("wild type") virus or bacteria (the one from the figure above) and will result in the measurable infection, and the second group will be the one infected with the mutant (figure below). This comparison is important because often the effect is not a complete absence of the infection, it might be reduced and then a possible quantification of the reduction of the infection due to the mutant compared to the wild type agent is the result of the experiment.