Mike the Mad Biologist

Why You Should Always Run Controls

One of the things that’s hammered into your head as a baby scientist is the importance of running controls. Typically, you run a positive control–a ‘gamed’ experiment where you know what the outcome should be and which tells you that the experiment is working–and a negative control which should not give any results at all (e.g., a PCR reaction without any DNA) to make sure there’s no contamination or other spurious results. It’s always puzzled me why people don’t like to run controls because if you don’t do the controls, you’ll have to redo the experiment, which is a lot more work than if you had done it the right way the first time. Sadly, some scientists, even once they’ve done grew up, still don’t do controls and somehow manage to get published.

In 2004, a group published a PCR survey that indicated that the TEM-1 beta-lactamase resistance gene was found in 91.3% of examined Streptococcus pneumoniae. This is a surprising result because beta-lactamases aren’t found in S. pneumoniae. In a recent paper in the Journal of Antimicrobial Chemotherapy, another research group found that when they ran only negative controls (i.e., just the reagents and no DNA), DNA contaminants yielded false positive results.

Oops.

This actually isn’t surprising, although it is astonishing that a paper could be published without adequate negative controls. When you run a PCR reaction, the enzyme that makes the whole replication procedure happen is known as Taq polymerase. Taq is short for Thermus aquaticus, which is the bacterium this enzyme was first isolated from. Many researchers often use ‘recombinant’ Taq–the gene is cloned into E. coli and then the Taq is produced by the E. coli*. Most people who work on E. coli (or genes found in E. coli) use ‘native’ Taq which is purified from T. aquaticus to avoid any possible contamination from E. coli DNA.

Here’s the really stoopid part: the cloning vector (the engineered mini-chromosome that has the Taq polymerase gene) also has a TEM-1 beta-lactamase gene because ampicillin resistance is used to kill off the bacteria that don’t have the cloning vector (and consequently can’t produce Taq polymerase). So if the manufacturer didn’t do a good job of eliminating DNA from the Taq polymerase, the negative control would indicate that there is a beta-lactamase gene even though there is no bacterial sample that is being tested.

Double oops.

Having said all of this, the rebuttal article begins on a very foul note (bold mine):

In 2004, an Asiatic group published an article entitled ‘Study on the molecular epidemiology of [beta]-lactamase TEM gene in isolated Streptococcus pneumoniae‘.

Asiatic? Sweet Baby Intelligent Designer, why not call them Mongoloid why they’re at it? You never hear someone rebut a paper by referring to a Jewish group**. Stupidity all the way around.

*People use recombinant Taq because it is either cheaper or included in ready-made mixes. T. aquaticus grows at very high temperatures and is far more expensive to culture than E. coli.

**On the other had, we Jews are fucking smart, so we likes our positive and negative controls.

Comments

  1. #1 Sandra Porter
    August 15, 2007

    I love it!

    Sometimes we find that people obtain perfectly good high quality DNA sequences from samples that didn’t contain DNA – i.e. empty capillaries.

    I am firm believer in lots and lots of controls!

  2. #2 Julie Stahlhut
    August 15, 2007

    I once photographed a gel that had perfect, absolutely gorgeous bands corresponding to a fragment of a mitochondrial gene for each of about 40 insects. I mean, it was so perfect that even the water blank yielded a strong positive. :-)

    A few more PCRs, controlling for every separate reagent in the cocktail, was all it took for me to trace this phenomenon to accidental contamination of my Taq polymerase with insect DNA. Don’t know how it happened, but it would have taken only a slip of the pipette tip. Needless to say, I threw the stuff out and ran the whole PCR over again.

  3. #3 michael Schmidt
    August 15, 2007

    There are a lot of biology experiments which are very amenable to simple controls; in other scientific disciplines, one has to be a little more creative in devising means of making sure you’re not fooling yourself. I, a chemist, often emphasize to biology students that controls are but a subset of the strategies used to make sure your inductive reasoning is robust. I have a biology master’s candidate working with me right now who was fixated on just finding positive and negative controls, but we had to proceed through a period of learning where we explored all sorts of other issues with the cutting-edge experiment we were trying to do. (Part of the problem was that the biology half of the project–the student and his advisors–seemed to think that what they were asking me to help the student on was well-established and routine, and not so cutting edge!)

  4. #4 Deborah
    August 15, 2007

    As a Medical Technologist (Clinical Laboratory Scientist), even the most automated commercial tests (well-established routine) require controls and troubleshooting in order to know whether your results are valid or not.

    Controls are just one indicator of whether you testing is correct. There are many variables in testing, from specimen integrity, pipetting, dilutions, instrumentation, reagent integrity, calculations and human error and so forth. It requires constant vigilance and troubleshooting in order to provide good test results.

    I would use a fresh pipette tip for every pipetting step unless there was no contamination issue (like pipetting diluent into clean tubes). Never pour back working aliquots back into the stock bottle. I would also be suspicious of any run that unexpectedly had the same result for all samples whether or not controls were run.

  5. #5 Mike Saelim
    August 15, 2007

    Doesn’t every scientific research group in the US have at least one part-Asian by ethnicity in it now?

  6. #6 DrYak
    August 16, 2007

    Well, at least in this case they weren’t the cause of a major scare, unlike the abysmal paper by Wakefield supposedly finding measles DNA in autistic bowel samples.

    For me the worst culprit is GFP. Once you have one plasmid with GFP floating around the lab, say goodbye to any chance of testing via pcr with GFP primers any other clones…

  7. #7 Bartholomew Cubbins
    August 16, 2007

    Nice post. Loved your take on their bigotry.

    /for those who homebrew Taq – there’s undoubtedly plasmid in your prep

  8. #8 PhysioProf
    August 16, 2007

    Controls can be a very intricate aspect of experimental design, particularly when you are establishing new methods. Personally, I think the casual use with which scientists throw around the terms “positive control” and “negative control” can be misleading to neophytes. This is because it can give the impression that there are only two controls needed for any experiment, “the positive control” and “the negative control”.

    In fact, as we all eventually learn, well-designed experiments will have numerous controls, each positive control designed to ensure that various aspects of the method are working as expected, and each negative control designed to rule out one or more potential artifactual reasons for a positive result.

    It is always fun for me to ask neophytes in my lab to explain their controls, when I know that some senior post-doc has just told them: “Run this condition as a positive control, and this condition as a negative control, and come back when you have results.”

    So I say, “Hmm, what is this condition for?” And they say, “It’s the negative control.” And then I say, “Hmm, that’s nice. What is it a control for?” And that starts them along the path to understanding of controls. It’s also fun to watch them come to the realization–sometimes painfully–that if they can only run a limited number of samples, negative controls are not the ones to omit.

  9. #9 Nick
    August 17, 2007

    re: “Asiatic”

    Look at the nationality of the authors. I would guess their use of “Asiatic” is due not to bigotry, but rather to imperfect translation and/or unfamiliarity with the negative connotations of particular words in English.

  10. #10 Suicyte
    August 17, 2007

    Please excuse my ignorance, but do you really suggest that describing someone as “Asian” is tantamount to calling him “Mongoloid” ? Or does the word “Asiatic” (certainly a strange translation from the Italian word) mean something very different in English? In any case, I find it a little rash to argue that if those guys don’t do proper controls, they are probably racists, too.

  11. #11 THobbes
    September 28, 2007

    In any case, I find it a little rash to argue that if those guys don’t do proper controls, they are probably racists, too.

    You seem to be misreading the post. The rebuttal article, not the original in which this PCR mistake was made, refers to “Asiatic” scientists. And yes, the use of the term “Asiatic” is completely unnecessary: for one, it’s incredibly vague (the scientists were Chinese, Japanese, Korean, etc., not “Asiatic”; would calling Kenyan researchers an “African” group tell us anything useful?), and second, the term “Asiatic” is strongly associated with discredited racialist theories, rather than geographic location. If the rebutting authors had to identify this group with such a term, “Asian” would have been geographically accurate.

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.