The Importance of Control Experiments

BIOC/MCB 568 -- Fall 2010

John W. Little--University of Arizona

An important part of most, if not all, experiments is a set of tests that are usually referred to as "controls". A control is a part of the experiment that tests whether the system behaves as it should. Imagine that you want to measure the rate of an enzyme reaction. This is done by using an assay for the appearance of the reaction products. Let's suppose that the starting substrate is colorless and the product is yellow; then the amount of yellow product should be proportional to the amount of enzyme added, or to the time the reaction is allowed to proceed. What controls would you do for this simple experiment? Think about this before you look below.

A reasonable expectation is that if you leave out the enzyme, no yellow color should appear. This is because the substrate is (we shall assume) stable, and should not break down to products over the time course of the assay. If it is unstable, one would be well-advised to find a better assay! So, in this experiment, the control is a separate incubation omitting enzyme. Perhaps another control would be to take a time point just after adding enzyme, to ensure that the enzyme itself doesn't contribute any yellow color; this might be overkill, if one could look at the enzyme and see that it is colorless.

Controls serve two somewhat related purposes in an experiment. The first is that they show that the system is behaving as one expects from previous knowledge, experience, and observation. This is necessary to allow an interpretation of the experiments. The second, which is equally useful, is that, if (or better, when) something goes wrong, it makes it possible to pinpoint the error in the experiment so that it can be corrected for the next time it is done.

A good example is a very simple kind of recombinant DNA experiment called a transformation. In this experiment, a recombinant DNA molecule called a "plasmid" is put into a bacterial host cell, using a procedure involving several steps outlined below. This plasmid carries a gene that makes the cell resistant to an antibiotic (say, ampicillin or tetracycline). The cells are normally sensitive to this antibiotic. Now, only a very small fraction of the cells take up the plasmid; for our purposes, let's say 1 cell in 10,000. The reason the antibiotic resistance gene is present on the plasmid is so that one can do what geneticists call a "selection" for those rare cell that carry the plasmid. In the presence of the antibiotic, only the cells with the plasmid can grow. So the procedure is the following:

1. Grow the cells in antibiotic-free medium. Wash them in a solution containing CaCl₂.

2. Mix the cells at 0 degrees C with plasmid DNA. Stand 0 C for a while, heat 2 min 37 C, chill.

3. Add growth medium, let the cells grow for 1 hr or so to express the antibiotic resistance gene.

4. Spread a portion of the cells on a Petri dish containing the antibiotic. Incubate the plates for 24 hr. Resistant cells grow, forming a bacterial colony.

The outcome of this experiment is that only those cells taking up the plasmid DNA should form a colony. So the simplest interpretation of finding colonies is that they arose from single cells that took up plasmid DNA.

Questions:

1. Assume you have done the above experiment. There are two likely outcomes. For each, give a hypothesis to explain the outcome:

a. You observe isolated colonies on your plate. Generally there would be a few to a few thousand colonies.



b. You see no colonies on the plate.



c. A less likely outcome is that you see a very large number of colonies, so many that they all grow together to make a confluent "lawn".



2. In each case, give other hypotheses which are also compatible with your observation. What controls can you imagine doing to rule out these alternative possible explanations, thereby providing support for your hypothesis?

a. A few hundred colonies could also mean:



b. Seeing no colonies could also mean:



c. Seeing confluent growth could also mean:



Plausible answers here.

Back to "How to read a paper"
Back to 568 home page