Moving at Warp Speed
Hi, my name is Molly Cinderella. This summer I’m working in Dr. Fred Nijhout’s lab. Most of my work will be focused on wing development in butterflies and moths. I’m going to be tracking development from caterpillar wing discs (called imaginal discs) to pupal wings to fully formed adult wings. The ultimate goal of my project is to determine how wings develop and where most of the growth occurs at different stages of development using image warping software and then looking at wings under a microscope to look for mitotic patterns that support the findings. If this research is successful in Precis and Manduca (the two species readily available in the lab), then the same techniques could be applied to several different species. It would be interesting to see whether or not insects in the same families shared similar wing growth patterns throughout development, but I’m not sure that there’s enough time to do that this summer because several different families of insects would have to be tested for that.
My first week in lab has been very productive. I’ve written a few different MATLAB programs that perform image warping and developed ways to test whether or not a particular warp is of high quality. I’ve learned a lot of new things in MATLAB so far - one of my side projects is to develop a GUI that runs the image warping software so that it can be operated without MATLAB. (Honestly, I’d love to have a copy of it on my computer. Before I received high-quality images to test the program on, I spent a very happy afternoon morphing people’s faces to Yoda. This helped me iron out several minor bugs in the code so that when I got usable images I was able to plug away.) The way this program works is that you take two similar images and establish several common landmarks between them. You then establish a grid on each of the images, and distort one image so that its landmarks correspond exactly to the same landmarks on the other image. The only real challenge is to pick appropriate landmarks -poorly chosen ones will lead to an unacceptable amount of distortion. In the later stages of wing development this isn’t too much of a problem because veins along wings establish really good landmarks, but in early-stage imaginal discs the veins are not yet fully formed, so choosing landmarks is more of a tedious trial-and-error process. (Pics next week, I know this is pretty hard to understand without them.)
So far I have crashed MATLAB at least twice doing this (I was so proud), something I never managed to accomplish while writing code for anything else. Manipulating images is fun but very hard on computers. The program takes quite awhile to run (especially if using more than a dozen or so reference points), so I usually have a book on either MATLAB or butterfly wing development open on my lap. In order to get high-quality images to use for the program (if the wing has been distorted at all during dissection or when transferred to a slide you end up with a lot of junk data), I had to learn how to do my own dissections. I started on Precis caterpillars because they are easier to dissect. I think I dissected about twenty on my first day just to practice. I actually kind of enjoyed dissecting the caterpillars - it required a lot of hand-eye coordination. The first caterpillar I attempted dissecting wasn’t sufficiently submerged in the saline solution (dissections are performed underwater because caterpillars dry out easily) after I knocked it out in CO2, so it started moving halfway through the dissection. I think I jumped about a foot in the air, and I’m pretty sure the lab tech started laughing at me. (She’s awesome, by the way). After that I spent a lot of time securing the caterpillars in place with the metal pins so that none of their spiracles ended up above water so that wouldn’t happen again. (It did, twice, but it wasn’t disturbing at all the second or third time) After that the insects were placed in the freezer. If all went well they never woke up and never felt anything, and putting them in the freezer was the gentlest way to kill them (they would have died soon after dissection anyway).
Next week I’m going to learn how to do dissections on Manduca and do several dissections of both species at the pupal stage. This is because Dr. Nijhout and I decided that the best way to approach the problem of creating these image warps throughout development is to start with more-developed wings and working our way backwards as far as we can, so getting good images at different stages in pupal development is very important. (That means I can’t poke holes with the forceps during dissection. I’m going to have to get much better at them). I’m really looking forward to getting some actual results next week now that my program is working well.
Some pictures:
.jpg)
Some Precis specimens. Not the ones I dissected.
This is the lab bench where I dissected the insects. Note the CO2 tank used to knock them out.
Someone is the lab is a poet, but no one knows who.
.jpg)
This is above the desk in the computer lab. It's where we our have lab meetings. I think it's beautiful.
Until next week,
Live long and prosper!
