Monday, February 11, 2013

Better measures of tiny bits of living jelly

My student and I had a paper accepted last week. I like this paper a lot. First some official details, then a short story about how the paper came to be.

Forthcoming in "Marine and Freshwater Research" 

The consistent, non-destructive measurement of small proteiform aquatic animals with application to the size and growth of hydra 

Daniel Levitis and Josephine Goldstein

Abstract: Hydra (Cnidaria), the basal metazoan most often studied in cellular, molecular and developmental biology, is difficult to measure because it is small, proteiform and aquatic. To facilitate broader organismal and ecological study of Hydra, we evaluate three methods whereby a polyp's body column can be measured by means of photomicroscopy. The volume, cylindrical surface area and surface area corrected for changes in body shape are all highly repeatable methods (r=0.97) when shape varies little. However, shape changes alter volume and cylindrical surface area. Repeated corrected surface area measures of the same individuals in markedly different positions yield standard deviations that are less than 5% of the mean measured area. This easy, non-lethal means of individual size measurement explicitly accounts for the flexible morphology of a polyp's hydrostatic skeleton. It therefore allows for the elucidation of how growth and size vary over time, age and food intake. We find that hydra change size dramatically day to day, and that while food level influences adult size, it has little effect on the early growth of recently detached buds. Finally, we discuss ecological and biological applications of this method. 

Part of why I like this paper so much is that my student did most of the hard parts, and the reviewers didn't ask for a lot of changes, and the editor (Dr. Russell Death) moved things along quickly and efficiently, so it is the first paper I've published that didn't feel like pulling my own teeth. It is a solid methods paper. It says, "here is a method for doing something that many scientists may want to do, that we didn't know a good way to do before."

In particular, it is a simple way to accurately and easily measure a tiny transparent aquatic animal with no hard parts or consistent shape without harming it.  Take a picture of it and with a bit of simple math calculate its surface area. Simple, elegant, inexpensive, non-invasive, biologically meaningful, everything I hoped it would be.

But the method presented in the paper is a substantially different, and better, method than the one we started out with. In fact, we were almost ready to submit the paper when we changed the method drastically.

I started out just wanting to measure a hydra polyp because it was necessary for another research project. Nobody had a method that worked halfway decently without killing the hydra. A hydra's body, although it shortens and lengthens, bends and twists is almost always roughly a deformed cylinder. So I said, "Hey Josi" (that's my student), why don't we develop a method to measure a hydra by photographing it and estimating its volume as though it was cylinder?" She agreed, and off we went, taking photos of hydra during our whole research project, for almost two years. We finished gathering our data, and found that indeed we could measure hydra this way. It didn't work great, but it kinda worked. You could use it to tell the difference between a really huge hydra and one that was just kind normal. Viola, unimpressive but probably publishable methods paper.

We wrote it up and were pretty close to submitting when I decided to see what hydra focussed papers had come out recently, and found that someone else had just published almost the exact same method in a nice paper with an interesting biological point. Our formulation of that method didn't work any better than hers did, and we had no point beyond "here is how to measure." We couldn't publish the same method again, even if we came up with it independently. After some cursing and self-recriminations for being so slow, I decided to see if it was possible to salvage anything of the methods paper. Josi took another set of photos and I used them to repeatedly adjust my formula, with limited biological reasoning, until something worked better than the formula just published. In fact, one formula I came up with by eyeballing my graphs worked much better than the method just published. So much better that the calculated value hardly changed at all even when the shape of the hydra changed drastically.

Now you are probably saying to yourself that this is blatant cheating. Trying different formulas (perhaps 100 of them) until something gives you the result you want is a pretty sure way to get the result you want, if you are persistent enough. But two things combined to make this not just okay, but beautiful. First, the formula I stumbled upon made obvious biological sense, even before I knew it worked. This formula represents a hydra as a roughly cylindrical bag whose skin stretches as it elongates itself, or folds and ripples as it contracts. In other words, it describes a hydra accurately and reveals something I didn't previously know about the way a hydra moves. Secondly, and as importantly, when I applied the formula to the main data set, which I didn't use to develop the formula, it still gave a highly consistent measurement for any individual, even as the shape of the individual changed. The method in fact works for completely different populations of hydra. You could take two hydra that looked the same size under the microscope and conclusively decide that one was bigger than the other. You could tell how much a young hydra grew each day. You could really measure the buggers, eliminating most of the noise inherent to previous methods.

Still more lovely, the method Josi and I had just developed could make use of the photos and measurements we had already taken, so redoing all our calculations and figures required Josi to write only a few extra lines of code (Thank you Josi, thank you R). We did a little rewriting to compare our method favorably to our other method, blamed attributed the other method on to the person who had just published it, made our biological argument to explain why the method worked, and we had a drastically improved paper ready to submit. There is a lesson in this somewhere about the scientific method.

1 comment:

jte said...

I don't get how measuring the critter's volume can tell you something interesting about its biology. What was it that you didn't know about hydra that you notice after applying the new formula?