Challenges of the Unseen World

Introduction

This course is made up of six challenges. Each challenge contains a different problem, one that you might encounter as a microbiologist and be asked to solve. To do so, you will draw upon different techniques learned during the course, obtain and analyze the data you need, and then present the solution to the class or to your instructor.

The Scientific Method

Scientists make new observations about the world and then provide an explanation for these observations. This sounds simple, but it isn't. For one thing, a new explanation must be viewed in the context of what has already been learned. Most often, the explanation is an extension or refinement of an earlier explanation. The new explanation is more powerful because it includes more observations, but is consistent with previous thinking. Occasionally, though, the new explanation is completely different from what was thought before. When that happens, it is an exciting moment in science.

How do we go from observation to explanation? The logical structure that scientists use, consciously or not, is called the scientific method, outlined informally in Fig. I-1. Scientists make careful observations and then identify those that need an explanation. They learn what is already known and then propose a hypothesis, a tentative explanation. The hypothesis must explain the new observation while being consistent with prior observations. In addition, it must be testable. This means that if the hypothesis is true, it will lead to predictions that can be tested by experiment. Scientists design and carry out these experiments and then ask whether the results match the predictions expected from the hypothesis. If they do not, the hypothesis is discarded and a new hypothesis accommodating these results is put forward.

Figure I-1 The steps in the scientific method.

The application of the scientific method as a series of steps is not always obvious from the course of scientific research and discovery. However, it still forms the logical underpinning of how scientists approach a problem. An example is the discovery that DNA is the carrier of genetic information (Fig. I-2). In 1928, Fred Griffiths, working with the bacterium Streptococcus pneumoniae, discovered that if you injected a mouse with dead cells of a virulent (disease-causing) strain, along with living cells of a strain that did not cause disease, the mouse developed an infection and died. By themselves, neither the dead virulent cells nor the living avirulent cells had this effect. Griffiths concluded that a "transforming principle" from the dead cells was converting the living cells to virulence. This was exciting because the acquired virulence was stably maintained as the cells grew and divided, indicating that the virulence trait was due to inherited genes. In other words, genetic information had passed from the dead cells to the living cells.

Figure I-2 Research leading to the discovery that the "transforming principle" is DNA.

What was the carrier of the genetic information? Since whole cells were used in the Griffiths experiment, there were many possibilities, but most of the bets were on proteins being the "transforming principle." The reason was that only proteins were thought to be sufficiently complex and various in their properties to convey genetic information. A critical new observation was provided by Dawson and Sia, who showed that the transforming substance could be extracted as a soluble component from the virulent cells and then used to transform the avirulent strain in a test tube. This meant that it might be possible to purify the transforming principle and determine its chemical properties, a fact that was recognized by Oswald Avery and his laboratory group. The first attempts at characterization indicated that it was not a protein; rather, the properties were consistent with deoxyribose nucleic acid, or DNA, another and surprising new observation. Avery and his colleagues set about testing the hypothesis that DNA was the transforming principle. If the hypothesis was true, then it would lead to several predictions that could be tested by experiment. In every case, the experimental results were consistent with Avery's hypothesis (Fig. I-2), resulting in a startling paper published by the Avery group in 1944. The idea that DNA was the carrier of genetic information was so unexpected that even Avery himself was reluctant to draw that conclusion, although, as you know, it has stood the test of time. A good review of this transformative moment in microbiology (no pun intended) is Cobb (2014).

There is an important but subtle logic behind scientific experiments. Science basically works by the process of elimination. Different hypotheses are tested by experiment and are discarded if the experimental results are inconsistent with the hypothesis. For example, the hypothesis that protein was transforming the cells was eliminated by the biochemical properties of the transforming principle. A hypothesis comes to be accepted when it is consistent with all the experimental results and when all other reasonable competing hypotheses have been ruled out by experimentation. "Reasonable competing hypotheses" depend on both our state of knowledge and our imagination. An awareness of this might have been one reason Avery was cautious about drawing the firm conclusion, in public at least, that DNA is the genetic material.

Experimental Design

From the foregoing it must be obvious that good experiments are the keystone of the scientific method. In designing an experiment, there are some things to keep in mind.

1. Does the experiment test the hypothesis? The purpose of a good experiment is to discriminate between hypotheses. Results that would be consistent with all the hypotheses under consideration do not help us to decide between them.

2. Is the experiment well controlled? Controlling all the possible variables except the condition you want to test is the best situation. In reality, this is not always possible, and there are often uncontrolled variables, variations in the experimental conditions in addition to what you want to test. Repeating the experiment multiple times, along with statistical methods, can sometimes be helpful when dealing with uncontrolled variables. However, statistical analysis cannot rescue experiments where the results are overwhelmingly influenced by the effects of uncontrolled experimental conditions.

3. Is the sample size large enough, and can the experiment be replicated? Sample size and replication of the experiment by yourself or others are closely related. Sometimes a result that seems to be real at first disappears upon replication. This is usually because the sample size was too small to begin with, and apparently real differences were just the result of chance. Suppose you hypothesize that a coin is weighted so that it will come up "heads" more often than "tails" after tossing. You decide to do the experiment of tossing the coin 8 times. If "heads" is the result 6 or more times, then you will conclude that your hypothesis is correct and you will publish your result. You get 7 heads and 1 tail during the toss, strong evidence, it seems, of a bad coin. However, while the probability of getting this particular result with a fair coin is only 3%, the probability of getting 6 heads or more is 14%. Your criterion for a bad coin would be met by a fair coin 14% of the time. The solution is to repeat the coin toss multiple times, which might seem obvious. However, many published experiments have not been sufficiently replicated. The inability to reproduce experimental results has become a major concern in the scientific community (Anonymous, 2016).

4. Are the accuracy and precision of your measurements adequate to support your conclusion? Accuracy refers to the closeness of a measurement to the true value, while precision refers to the reproducibility of a measurement: how often repeated measurements will give the same value. Both must be taken into consideration when drawing conclusions from an experiment. For example, a small but real change due to different experimental conditions might not be detected if the measurements are inaccurate. Imprecise measurements, on the other hand, could result in the real change becoming obscured by the random "scatter" of different data points.

5. Could observer bias influence the results? When scientists do experiments, they often have a desired result in mind, usually the one that supports their favorite hypothesis. This can lead to the unrecognized manipulation of results to favor this hypothesis. Sometimes rationalizations like "This value is much smaller than the rest: obviously there was a procedural error so it should be discarded" are used as a justification. This is a particular problem with students in lab classes. Often they think they know the expected outcome of an experiment. If some measurements do not support this result, they immediately assume that these were due to experimental errors and can be discarded.

For practice, consider the following situation:

A marine microbiologist suspects that iron in seawater stimulates the activity of a particular enzyme in the microbe she is studying. She takes eight samples of the seawater over the course of a month and adds the same amount of bacteria to each when she is ready to do the experiment. To four of these she also adds iron...