Errol Morris has this “Op-Doc” (Opinion-Documentary, as odd as that sounds) over at the New York Times. Technically, it’s about historical research, but I think the phenomenon holds true for biology, as well. As interviewee Tink Thompson puts it, “if you put any event under the microscope, you will find a whole dimension of completely weird, incredible things going on”. I’ve always loved Morris’ work. A lot of it deals with issues of knowing—how we know what we know, how we sometimes deceive ourselves, what evidence does or doesn’t say. Important issues for any scientist.
I finally got around to seeing the film Contagion, a realistic portrayal of what a serious viral pandemic would look like in our day and age. Unlike the vast majority of outbreak films, Contagion gets a lot of the science right. The basic reproductive number R0, for example, makes an appearance in the film. The film’s fictional MEV-1 virus was inspired by Nipahvirus. MEV-1 has an R0 of 2-4 and a mortality rate of ~25% — severe, but realistic. Epidemiologist Ian Lipkin consulted on the film.
My favorite line is when the military asks CDC Deputy Director Lawrence Fishburne if the virus was a bioterrorist attack, like a weaponized bird flu. Fishburne reponds, “No one has to weaponize the bird flu. The birds are doing that.”
One of the things that seems wierd to me about science as a profession is how the job often requires both esoteric activities like probability theory or genetic engineering and mundane menial labor like washing dishes or mixing large amounts of dirt. As a graduate student I used to want to do everything in the lab myself: making media, counting plates, all of it. Part of it was that I wanted to know how every part of my experiments worked, and part of it was perfectionism and paranoia. Now, I find myself wanting to teach someone how to do that stuff well, and have them do the tedious work for me. Especially the part where you’ve already done the pilot experiments and the first replicate so you already know what the answer’s going to be — you’re just getting clean data for the paper. Now, I’d rather develop the assays, get them working reliably, and then let someone else finish out the data collection.
I have to say, though: my experience of lab work has recently been completely transformed since purchasing an mp3 player (I’m a slow adopter). Before, it was like, “I need to make media? Again? Grumble grumble…” Now it’s like, “Aw, hell yeah”.
Dictyostelium and its relatives are getting some time in the limelight from Carl Zimmer in the New York Times. The article even includes some coverage of our lab’s work on the evolution of cooperation.
Last week I had the opportunity to visit the University of Houston. When our lab was still at Rice I went to a talk or two at UH but never really got much chance to interact with the research groups there. So it was great fun to talk evolutionary genetics with Tim Cooper and his lab, social amoebae with Elizabeth Ostrowski, evolutionary networks with Ricardo Azevedo, behavior and morphology with Tony Frankino, ant behavior with Blaine Cole, and food and art with Dan Graur (over dinner).
The audience for my talk was mostly biologists, not necessarily in my field of study, many of whom may not have been ecologists or evolutionary biologists. So I used the opportunity to develop what will hopefully become my job talk. I’ve been inspired by Will Ratcliff’s “Morgan Freeman” philosophy for scientific talks. The idea is that it should be a story, presented with engaging images, and it’s your job to guide the audience through that story with an easy-to-understand narrative.
One thing I’ve learned: making slides is no substitute for actually practicing a talk. Even if you mentally run through your narrative as you make slides, when you actually get to talking you’ll inevitably discover places where the narrative isn’t as clear as you thought it was.
smith j (2011) Distinguishing causes of virulence evolution: Reply to Alizon and Michalakis. Evolution In press. doi:10.1111/j.1558-5646.2011.01428.x | Journal
Abstract: In a recent study of the symbiosis between bacteria and plasmids, the available evidence suggests that experimental evolution of plasmid virulence was primarily driven by within-host competition caused by superinfection. The data do not exclude the possibility, however, that a trade-off between virulence and infectious transmission to uninfected bacteria also played a minor role.
This one’s an in-print discussion between Samuel Alizon, Yannis Michalakis, and myself. They were worried that some researchers might interpret my earlier paper about plasmid evolution as rejecting the hypothesis that pathogen virulence can be influenced by a trade-off between infectious transmission and virulence. So they made some mathematical models to determine what kinds of evidence we’d need to really rule out the influence of such a trade-off. The work I did doesn’t rule out a trade-off, but it does show that competition among pathogens within hosts was necessary and sufficient to explain how plasmid virulence evolved.
I found it nice (and useful) to talk with the authors before we sent in our respective manuscripts. It helped clear up some places where we were using language in a way that could be misinterpreted and to focus on places where there might be real issues. In the end, my impression is that we basically agree about what my paper does and doesn’t show, but they wanted to use it as a sounding board for how we should go about testing virulence theory.
Scott M, Gunderson CW, Mateescu EM, Zhang Z, Hwa T (2010) Interdependence of cell growth and gene expression: origins and consequences. Science 330: 1099-1102. Journal | PubMed
Using extremely simple models of cell physiology, Scott and colleagues are able to predict how unnecessary gene expression reduces bacterial fitness, how growth rate changes due to nutrient quality, and how growth rate is reduced by sublethal concentrations of antibiotics that inhibit translation. They also derive the classic Michaelis-Menten equation for growth rate as a function of resource concentration. Their models assume that growth is linearly proportional to translation rate and that there are three classes of proteins: those used to acquire nutrients, those used to make more protein, and a class that is unaffected by nutrient availability. With this, tons of results pretty much just fall out. The paper is a great example of how systems biology links molecular details to whole-cell phenotypes.
“Like Ohm’s law, which greatly expedited the design of electrical circuits well before electricity was understood microscopically, the empirical correlations described here may be viewed as microbial ‘growth laws,’ the use of which may facilitate our understanding of the operation and design of complex biological systems well before all the underlying regulatory circuits are elucidated at the molecular level.”
Hopefully the amoebae will act more or less the same as in our previous lab, but I wouldn’t bet too much on it. My experience in Greg Velicer’s lab showed that sometimes experiments can give different results in different labs. Paul Turner found that the culprit is often the water. Knowing this, I spent many late nights back in Houston finishing up a big block of experiments to get a discrete chunk of data collected in only one lab. I’ve yet to analyze it all, but I’ll probably need to do a couple follow-up experiments to get at the cause(s) of the effects there. Hopefully any differences will be minor.
I also want to capitalize on Dicty’s photogenicity (by microbial standards) and get a bunch of cool pictures to use in talks, on this site, and maybe even as cover images for journals.
When I originally got interested in microbes, it was mostly for somewhat obscure academic questions about mobile genetic elements: how does evolution solve conflicts to create cohesive individuals instead of loose associations of genes? The abundance of plasmids and phage, and the fact that they sometimes carry genes for antibiotic resistance or pathogen virulence, seemed like a situation where that integration was not fully complete, which made them a good system to study. The fact that microbes and their mobile elements had significant effects on human health was a nice bonus, but not my main motivation. Mostly, I was happy that a lot of the genetic details had already been worked out. I liked the infectious disease aspect, but again mainly for academic reasons about conflict evolution, not because of its practical importance for humans.
Now, I find myself finding the applied aspects more compelling. They’re still not the primary thing, but if you have a choice of systems to study for academic reasons, why not study one that’s also relevant to human welfare? I’ve spent the last several years studying microbial cooperation in Myxococcus bacteria and Dictyostelium amoebae. These organisms are pretty cool biologically, but they really don’t have much to do with humans. Other examples of microbial cooperation are maybe more plain but also more relevant, like the Pseudomonas bacteria that kill people with cystic fibrosis.
I also recently realized the potential practical implications of my old grad school work on plasmids. In those experiments, I observed a rapid, repeatable loss of antibiotic resistance and a suppressed proliferation of bacteria — both of which are desirable clinical outcomes. I’d never really thought about those results from an applied standpoint. Why did my plasmids evolve to get rid of resistance genes while in other people’s experiments they stuck around even when there were no antibiotics? Is it possible to influence natural plasmid evolution to follow the path I saw? It’s definitely worth following up on.
In my field, data is often some quantitative measure of abundance: How many cells are there? What fraction are this genotype or that genotype? How does the number or fraction change over time? Typically, the raw experimental measurements get put into a spreadsheet, analyzed, and then that’s what gets put into figures or statistics programs.
A lot can go on in those spreadsheets, and a lot of mistakes can be made, yet in my experience there’s rarely any discussion or training about what happens to data between the experiment and the paper. At no time in grad school or either of my postdocs did anyone ask to talk to me about my spreadsheets. Over the years, I’ve figured a few things out and caught a lot of potential mistakes, so there are definitely things to discuss. I’ve also noticed some issues the few times I’ve looked closely at other researcher’s spreadsheets. It seems really strange to me that labs can spend so much time getting the experiments right but then be kind of careless with the data.
If you were going to teach a new grad student best practice for handling data and calculations, what would you cover? Here are some things that come to mind:
Keeping the raw measurements collected in a single place for easy access later on
Data quality control: making sure you didn’t type in the wrong number, or type it into the wrong place
Annotating data so that you can go back to your lab notebook or original files to verify specific entries
Identifying bad data points that should not be included
When you should average multiple measurements, and how you should do it. Whether you should take the arithmetic or geometric mean, for example.
Making sure that lists of calculated values use the same formula for every entry
smith j (2011) Superinfection drives virulence evolution in experimental populations of bacteria and plasmids. Evolution 65: 831-841. Journal | PubMed | Faculty of 1000
Abstract: A prominent hypothesis proposes that pathogen virulence evolves in large part due to a trade-off between infectiousness and damage to hosts. Other explanations emphasize how virulence evolves in response to competition among pathogens within hosts. Given the proliferation of theoretical possibilities, what best predicts how virulence evolves in real biological systems? Here, I show that virulence evolution in experimental populations of bacteria and self-transmissible plasmids is best explained by within-host competition. Plasmids evolved to severely reduce the ﬁtness of their hosts even in the absence of uninfected cells. This result is inconsistent with the trade-off hypothesis, which predicts that under these conditions vertically transmitted pathogens would evolve to be less virulent. Plasmid virulence was strongly correlated with the ability to superinfect cells containing competing plasmid genotypes, suggesting a key role for within-host competition. When virulent genotypes became common, hosts evolved resistance to plasmid infection. These results show that the trade-off hypothesis can incorrectly predict virulence evolution when within-host interactions are neglected. They also show that symbioses between bacteria and plasmids can evolve to be surprisingly antagonistic.
Some choice nuggets I came across recently in a short perspective by Nadell and colleagues at Princeton:
A recent paper in BMC Biology tests central elements of [social evolution] theory by manipulating a simple bacterial experimental system. This approach is useful for assessing the principles of social evolution, but we argue that more effort must be invested in the inverse problem: using social evolution theory to understand the lives of bacteria.
The interaction between social evolution theory and microbiology holds enormous potential for enriching our knowledge of bacterial behavior, but to realize this potential we must ensure that information flows in both directions between these formerly disparate fields. At present, social evolution theory has benefited from simple experiments with bacteria, but microbiology has not equally benefited from social evolution theory.
Pretty much sums up why I find some microbial cooperation papers really cool but others not so much.
Nadell CD, Bassler BL, Levin SA (2008) Observing bacteria through the lens of social evolution. Journal of Biology 7:27. (doi:10.1186/jbiol87) http://jbiol.com/content/7/7/27
Last summer, Nowak and colleagues published a perspective in Nature arguing against both the factual correctness and the scientific usefulness of kin selection theory. It rubbed a lot of people the wrong way. The paper has a lot of problems, but I’ll leave most of that argumenttoothers. Instead, here’s some correspondence I sent to Nature that they declined to publish:
Nowak and coauthors (Nature 466, 1057-1062; 2010) claim that kin selection theory is limited by its inability to describe evolutionary dynamics and its requirement for weak selection, pairwise interactions, additive fitness effects, and specific kinds of population structure. Shortly before this article was published, my colleagues and I published an extension of kin selection theory that overcomes all of these issues (Science 328, 1700-1703; 2010). We agreed with Nowak and colleagues that many kin selection models are hard to apply to real data, but we took a different approach. Identifying and solving problems is a productive strategy of scientific inquiry; wholesale dismissal of an active research program is likely to generate more heat than light.
I think there are issues with kin selection theory very much along the lines that Nowak and company point out. But it’s not so much that they make the theory measurably wrong (on this point I disagree with the Nowak paper), but that they make it less useful, hard to apply, and easy to get wrong.