Tuesday, October 28, 2008

Definitions Matter: Negative Selection and Postive Selection

In recent issues of the Proceedings of the National Academy of Sciences (USA) we have an interesting example of misuse of a key term in evolution.

The paper in question is by Sun et al. (2008a) of the University of California, San Francisco. The title of the paper is important: "Experimental evidence for negative selection in the evolution of a Yersinia pestis pseudogene." Here's how they describe this negative selection in the abstract,
Yersinia pestis, the agent of bubonic plague, evolved from the enteric pathogen Yersinia pseudotuberculosis within the past 20,000 years. Because ancestor and descendant both exist, it is possible to infer steps in molecular evolution by direct experimental approaches. The Y. pestis life cycle includes establishment of a biofilm within its vector, the flea. Although Y. pseudotuberculosis makes biofilms in other environments, it fails to do so in the insect. We show that rcsA, a negative regulator of biofilms that is functional in Y. pseudotuberculosis, is a pseudogene in Y. pestis. Replacement of the pseudogene with the functional Y. pseudotuberculosis rcsA allele strongly represses biofilm formation and essentially abolishes flea biofilms. The conversion of rcsA to a pseudogene during Y. pestis evolution, therefore, was a case of negative selection rather than neutral genetic drift.
Hmmm ... something about this form of "negative selection" seems puzzling. Does anyone see what it is?

The article was published online on June 3, 2008 and appeared in the June 10, 2008 issue of PNAS. It was communicated by National Academy member Stanley Falkow of Stanford University.

In this week's issue (Oct. 21, 2008) we see a letter from Jianzhi Zhang of the Department of Ecology and Evolutionary Biology, University of Michigan (Zhang, 2008).
There are two types of natural selection in biological evolution: Positive (Darwinian) selection promotes the spread of beneficial alleles, and negative (or purifying) selection hinders the spread of deleterious alleles (1). Pseudogenization is normally detrimental and prevented by negative selection. However, changes in genetic background or environment may render a formerly useful gene worthless, leading to the relaxation of the negative selection. Consequently, mutations disrupting the gene are fixed by genetic drift, and the gene becomes a pseudogene. This is the common type of pseudogenization by neutral evolution. Sometimes, however, a previously useful gene may become harmful to an organism. In this case, mutations destroying the gene would be beneficial and would be fixed by positive selection. Thus, pseudogenization can be adaptive (2). Recently, Sun et al. (3) reported an excellent example of adaptive pseudogenization, convincingly demonstrating that gene loss can also serve as an “engine” of evolution (4). Nevertheless, instead of calling it “positive selection,” they mistakenly used “negative selection.” The case involves Yersinia pestis, the agent of bubonic plague that is frequently transmitted by fleas. The authors found that the rcsA gene of Y. pestis became a pseudogene in the last 20,000 years (3). Replacing the rcsA pseudogene with its functional version represses the formation of biofilms in fleas (3), which would reduce the transmission rate of the bacteria. That is, the pseudogenization of rcsA allowed the formation of Y. pestis biofilms, which enhances the transmission of the bacteria, and hence was likely driven by positive selection.
That looks like a pretty devastating criticism to me. I'm convinced that the title of the paper was inaccurate. They were publishing an example of positive selection and not negative selection as claimed.

The authors replied in the same issue (Sun et al., 2008b).
In our article (1) we used “negative selection” to succinctly convey that a previously functional allele became deleterious and therefore was removed by natural selection. However, Zhang (2) is correct that our usage was contrary to the usual meaning. Olson's term, “adaptive gene loss” (3), would have been more appropriate. We are gratified that Zhang agrees with our conclusion that the pseudogenization of rcsA was adaptive.
Translation: "We really screwed up."

How did this happen? Normally, before a paper is published the work is presented at meetings and in lab group meetings. Was there nobody who recognized that the authors were using the wrong term? Clearly the authors themselves (all three) never questioned what they were putting into the title. Clearly the person who communicated the article didn't either, and neither did any of the reviewers.

What's happening to science these days? Now, don't get me wrong. These sorts of things happened in the "olden days" as well but I'm convinced that the problem is much more serious today. There is too much stuff being published that should never have made it past the lab group, let alone past reviewers.

Here's a question for everyone who has read this far. What should be done with the original paper? The title is wrong. How do we alert people to the fact that the authors have agreed that they made an error?

Sun, Y-C., Hinnebusch, J.B. and Darby, C. (2008a) Experimental evidence for negative selection in the evolution of a Yersinia pestis pseudogene. Proc. Natl. Acad. Sci. (USA) 105:8097-8101. [DOI: 10.1073/pnas.0803525105]

Sun, Y-C., Hinnebusch, J.B. and Darby, C. (2008b) Reply to Zhang: Adaptive gene loss in Yersinia pestis rcsA pseudogenization. Proc. Natl. Acad. Sci. (USA) 105:E70; published ahead of print October 15, 2008. [doi:10.1073/pnas.0807434105]

Zhang, J. (2008) Positive selection, not negative selection, in the pseudogenization of rcsA in Yersinia pestis. Proc. Natl. Acad. Sci. (USA) 105:E69; published ahead of print October 15, 2008 [doi:10.1073/pnas.0806419105]


  1. You exaggerate to shred your inner garments over the state of "modern science".

    Why? First, the work describes an interesting case well enough, which remains interesting, regardless of what yo may call it.

    Second, because "positive selection" IS a murky an largely abused notion. I think there is an epistemological problem there, such that almost anything can end up being described as "positive selection" ultimately based on good ole adaptationist thinking and "darwinistic" bias.

    For instance, when we call this "positive selection", we imagine a scenario where a given mutant bacterium gets and advantage in the population; it is "better", outcompeting other less fit bacteria. However, this scenario requires us to imagine a
    population that, despite living in fleas, cannot infect that good because it carries a harmful gene; in other words, we must assume the population goes through an initial "live like shit" phase. Weird, huh?

    Is this realistic? Do these people even conceive of the possiblity that the entire initial population is NOT living like shit, but in some other lifestyle? Film-making mutants have the possibility of colonizing new environments, such as fleas. In this scenario, the pseudogenizing mutation is initially neutral; it allows colonization of the flea, which is like a founder effect. Colonizing a new environment is quite a different thing from outcompeting the population of origin... Further, in this new environment, going back to the functional mutation is, without doubt, negatively selected.

    So, competitive positive selection? or drift, colonization and negative selection? At least to me this is not a matter of simply applying definitions as if these were delivered by nature itself. I think there is a genuine problem here.

  2. This comment has been removed by the author.

  3. There are links to the commentary on the PNAS online version of the article.
    As long as people are clear on what the authors meant, is their error really that important?

  4. Would it be possible to change the title? "Experimental evidence for positive selection (formerly negative selecton)"?

  5. Devin asks,

    As long as people are clear on what the authors meant, is their error really that important?

    Let's say you're looking for good examples of positive selection in natural populations of bacteria.

    Would you find this article in a PubMed search?

    I think it matters on many levels. We try to teach our students how to think critically and part of that process means defining your terms. This paper seems to be an example of very fuzzy thinking.

  6. The term "positive selection" is a mushy term, so it's funny that Larry would complains about it as if it were carved into stone, with a precise meaning . No sir.

    For instance, "positive selection" has been used for sequences that have a greater rate of translated vs neutral (3rd codon) substitutions.

    Or, it is used to connote a "creative" or "shaping" role of selection (rather than simply making an available trait more frequent).

  7. The title can certainly be changed, as an erratum. It happened to a colleague of mine in the same journal:

    Du W, Maniatis T. "An ATF/CREB binding site is required for virus induction of the human interferon beta gene [corrected]" Proc Natl Acad Sci USA. 1992 Mar 15;89(6):2150-4. Erratum in: Proc Natl Acad Sci USA 1992 Jun 15;89(12):5700.

  8. The term "positive selection" is a mushy term

    Nothing mushy about it. Or, to be super precise, it is as much mushy as any perfectly common term. That some people who have no clue use it loosely does not make it mushy (cf. "liberal").

    Increase in frequency = positive selection. Decrease = negative. Anything more simple than this?

  9. Woa, a single post evoking a panoply of problems in modern biology.

    People doing useful and successful science without understanding the basic tenets of the subject? - Check.

    Publishing in PNAS the old corrupt way? - Check.

    Failure of peer review? - Check.

    Too much bullshit published and going unnoticed? - Check.

    The answer to the question Larry posed in the end:

    PNAS and the authors have to have the guts to retract the paper and publish a new one with the basic error corrected. An erratum or corrigendum won't cut it in such a case.


  10. "Increase in frequency = positive selection. Anything more simple than this?"


    Given you made such simple requirements a gene that increases its frequency by a foundation effect or a bottleneck fully qualifies as an example of "positive selection".

    Where's the good ole darwinian "betterment" there? Where's the competition?

    Seems like you'll have to elaborate.
    Tip: there's no way around discussing a correspondence of phenotype to environment.

    I insist: What has happened in the evolution of Yersinia is not at all necessarily best described as "positive selection".

    Indeed, I find the argument to be typical: Point your finger to any adaptation and just say "positive selection". Works like a charm.

  11. Since it is online, have the manuscript corrected noting the past version, the error and the letter exchange. This can be denoted as well in the PDF version. The error is corrected and a constant reminder is in place for the editor, reviewers and authors of their serious error.

  12. DK says,

    Increase in frequency = positive selection. Decrease = negative. Anything more simple than this

    An allele could increase in frequency in a population by random genetic drift. Similarly, an allele can decrease by random genetic drift.

    Is that what you call "simple"?

    Your definitions are incorrect.

  13. To a. vargas and Larry:

    Of course not. What I wrote refers to SELECTION. So IF there is a selection, the direction is straightforward. That's what I meant. In the case of Y. pestis, the mutation is clearly adaptive, not neutral.

  14. So, if I understand, in your definition positive selection = adaptation, and vice-versa.

    What do you think about my argument of a mutation that colonizes a new environment without outcompeting its population of origin?

    Adaptationists are usually strangely deaf to that argument

  15. You need to read up a little on definitions of positive selection. Some authors argue that positive selection "creates the fit" while negative selection "eliminates the unfit"
    What do you think of that?

  16. I don't want to read a little on how others define positive selection. Here is my definition of positive selection:

    It is what happens in nature when a trait confers increase in fitness.

    Saying that "positive selection "creates the fit" is a vague statement that is of no value to me.

  17. This comment has been removed by the author.

  18. "Saying that "positive selection "creates the fit" is a vague statement that is of no value to me"

    Well, you should care, because it is still quite mainstream to say that negative selection merely eliminates the unfit, whereas it is positive selection that has an active role in the origin of adaptations. This is my main beef with people ts tossing around the notion of "positive selection"

    "It is what happens in nature when a trait confers increase in fitness"

    That's a fairly good definition, but keep in mind that fitness is relative to the rest of the population.

    So, when a new mutation colonizes a new environment, it founds a new population without outcompeting its population of origin. In this new, growing population everyone is as fit as the next.

    Is that positive selection?

  19. So, when a new mutation colonizes a new environment, it founds a new population without outcompeting its population of origin. In this new, growing population everyone is as fit as the next.

    Is that positive selection?

    If that new mutation ALLOWED colonization of the new environment then yes, it is undoubtedly a positive selection. More places to propagate = higher fitness.

  20. If you are going to use the word "selection" then you have to talk about a "selective pressure" affecting different phenotypes in the same population.
    What is the selective pressure in that example? Is selection being applied to a single population, if part of the population cannot even live in the same environment?

    I cannot agree with you calling that "selection".

    If anything, the mutation is selecting the environment, and not the other way round.

    Another question. Are you aware that most people would call the peppered moth case a simple case of negative selection (elimination of the white moths)

    To call it "positive selection", selection should have a role in the generation of a new phenotype, not just the elimination of another.

    But you would call that "positive selection for black moths".

    In fact, in your book, negative selection always implies there is "positive selection"....of whoever is not being eliminated. Hehe. You can ALWAYS find positive selection!!!


  21. Sure as hell it's still a selection. The other guys couldn't live in the new environment because of the negative selection pressure exerted on them by that environment (otherwise they would be living there, a no brainer!)

    So the mutated guys who managed to survive there were positively selected.

  22. so, we are under negative selection for nay environment we don't live in?
    I am under negative selection for living in water, for instance.


    Take care now. Byebye then

  23. Yes. Finally you understood it. Yes, you are not living in the water because there will be a very strong negative selection against your there. And you are not living on another planet that is just like Earth because the environment exerts a strong negative selection against you getting there - you are not adapted enough to reach that planet to begin with.

  24. If two beneficial alleles compete, which one would be positive or negative? Is it a relative term?