The Futile Cycle

It’s 12 hours till I finish writing my NSF fellowship application, reading two for tomorrow’s class, and analyze my data for tomorrow’s group meeting. I’ve got a full cup of coffee, half a bag of chocolate, it’s dark, and I’m wearing sunglasses.

Hit it.

This past week was “Fall Break”, which I’d not experienced until now (my college had no such things…). My department likes to ease first years into the flow of things by massive bribery with tons of free food at colloquiums, talks, and events for the first years.

Unfortunately, that all went away during fall break, such that I was left to my own devices. My supply of ramen was non-existent (I know, should have planned better), so I had to look elsewhere. And, lacking a car and a kitchen, ’twas slim pickin’s.

So when the dining halls reopened here, my reaction was rather like this:

This week I will most likely not be posting, mainly due to my taking my “Molecular Biology of Prokaryotes” exam. It reminds me of the theoretical mathematics class I took as an undergrad; most of my time was spent staring at an empty page, waiting for inspiration to hit (or as Douglas Adams puts it, “until your forehead bleeds”).

As one of the professors here said, “When I read through the midterm, it was not obvious to me how to solve the problems.”

This PhD Comic hits too close to home, especially since I’m applying for an NSF fellowship right now. For the planned research summary, I have to supply keywords, background, detailed methodology, anticipated results, broader impacts, and citations and references all in two pages of 12 pt. Times New Roman font, with 1 inch margins.

But since I’m begging for money, I guess I can’t really complain too much. I’ve learned a lot of tricks for beating the page limit, though. Hyphenating long words at the ends of lines is really key; it can save you several lines of text. The most painful thing to include in the proposal is a citation to a book with a really long title, because there’s really no way to get around putting the entire title in the citation. The long title, however, almost always wraps the text to the next line, taking up two whole lines…

One of the amusing things to look at when I have time in between experiments is to look at what kind of search phrases people use to find the site. The best ones are from the long tail of search phrases that only one person would ever type:

“earthworm hammer dissection”. I don’t know about you, but generally I don’t use hammers when dissecting things as small as earthworms…

“vitameatavegamin medicine”. I don’t really want to know how many people are taking medical advice from “I Love Lucy”.

“is well done steak higher in water than rare”. I’m not really sure what this means. Are they asking whether well done steak has more water? (It doesn’t.) Are they asking whether it floats compared to rare? (I have no idea.)

“how in the hell does a person create a ‘table of contents’ in microsoft word 2007″. Someone is really frustrated! I don’t know. I’ve sworn off using Word unless I absolutely have to.

“should we treat hospitals differently than we treat farms, car”. There are lots of things wrong with the current health system, but that doesn’t mean we should take it out to the pasture. (cricket) Is this asking about subsidies? I hope we treat our hospitals differently from cars…

“when i hear biochemistry, i think of”. Well, I think of lots of pipetting. And blots. Lots of blots, lots of buffer. And sometimes some nifty, but really expensive, machines.

“what is a parafilm in chemistry?” I’m a big fan of parafilm. It’s a form of wax in a sheet, with some additives that make it more elastic. It’s like the sophisticated man’s plastic wrap, the scientist’s duct tape. More praises cannot be poured over parafilm.

After the first week of classes in graduate school, I’m a bit wrung out. I’ve taken graduate classes before, but not too many focused on paper-reading; the other classes had maybe two or three papers per week, but the current course I’m taking assigns two or three papers per class, which means around 10 hours of reading a week, assuming a little more than an hour per paper (we have to read them inside and out, understanding them in deep, experimental detail). With having around 6 or 7 seminar talks a week, two classes, and lab work to be done, and fellowships to write, I’m glad the weekend is here.

On the other hand, though, all the classwork means that I’ve been getting a lot of exposure to some very classic papers. The “fluctuation test” paper by Luria and Delbruck, for example, is from 1943, ten years before the structure of DNA was deciphered by Watson and Crick. Luria and Delbruck try to figure out whether bacteria become immune to viruses by mutations or because by chance some survive and acquire an immunity which is heritable, and they use some pretty clever math to do it.

The basic idea they use is that of the “jackpot”. Imagine that you’re playing a slot machine, and you win very rarely, but when you do, it’s a huge, huge payout: $100 million. Let’s say you pull the lever a million times, enough that you have some chance to actually win once or twice, but not enough that you’re sure to win. Now, let’s say that you have 20 people who all go to this machine and pull the lever a million times. Because one person might win three times, and another might not win at all, each person’s winnings will vary a huge amount from the others’. That’s the “jackpot” idea, that small differences get amplified a whole lot, because the payout is huge.

Now back to bacteria. There are two possible ideas, that bacteria all have a small chance of surviving the virus randomly, but that once they survive they’re immune (and pass that immunity on to their children), or that a small fraction of bacteria have a mutation that makes them immune, but most bacteria are susceptible.

The thing is, mutations are kind of like jackpots. If you start with one cell and it divides, acquiring mutations along the way, then in the early stages where there still aren’t that many cells, one cell might get a mutation. Since bacteria multiply quickly, that cell will then have lots of descendents that are also mutated, and so there’s a huge “payoff” for having an early mutation, because that change gets amplified. The number of mutations early on is random, so in two different experiments, there’s a good chance that the number of mutations you find in one is very different from the number in the other.

On the other hand, in the surviving and adapting viewpoint, there’s a low chance of surviving no matter how you go about it, and that characteristic doesn’t get amplified when bacteria have descendents. There’s no “jackpot” effect, and so the level of immune bacteria wouldn’t vary as much from experiment to experiment.

So, Luria and Delbruck did the experiment a lot of times. They took a small number of bacteria, grew them up, and tested them for immunity against a virus; and they saw the jackpot effect. Thus, mutations are the underlying cause for the resistance to virus. Bingo!

There are some people around who say “nuke-ular” (as in, “NOOK-yoo-ler”) instead of the conventional (and correct) “nuclear” (”NOOK-lee-er”). I can accept that this is a relatively easy thing to fall into, since “nuke” is a fairly common slang word, and both words have something to do with nuclear power, nuclear weapons, and so on.

Today, a professor was giving an introductory lecture, and I was half-paying attention (I’d heard this part before). But wait, did she just say “nuke-ulotide”?? Yes, yes she did. Not “nucleotide” (NOOK-lee-oh-tide), but “nuke-ulotide” (NOOK-yoo-lo-tide). Since this was a bioinformatics course, I spent the rest of the lecture trying to stop the bleeding from my ears.

The entire department went on a retreat this past weekend, which was a grueling but pretty wonderful experience. We have a rather large department (arond 50 labs, spread across vast swaths of biology, including microbiology, cell biology, development, neurology, virology, cancer research, metabolomics, mathematical biology, bioinformatics, systems biology, structural biology, biochemistry, infectious disease, and so on), so trying to get an idea of what’s going on here is a rather Herculean task; the retreat not only allows us first years to get a sense for all the labs and talk to faculty and older students, but it also allowed lots of grad students and professors to talk about research all day and form new collaborations. It almost seems like a third of all the research presented during the retreat was done in collaboration with another lab. It was a great thing to have almost as soon as I got to campus. I bonded with other grad students, met almost everyone in the department that was there, found other labs that were doing great research, and learned a heck of a lot of biology.

We invited a great guest speaker to the retreat as well, Charles Ofria, a computer scientist who researches the evolution of “digital organisms” — computer programs that can self replicate, compete, and mutate. He started off with the story of Core Wars, in which programmers wrote programs to fight each other for control of a computer, to Tierra, where programs could mutate and evolve, to his Avida system, which allowed Ofria to control and analyse almost any aspect of the evolution of the organisms. He was able to look at a large variety of theoretical problems, including the evolution of complex traits (e.g. critically examining the arguments of Behe’s “irreducible complexity”), fitness changes in a rugged landscape, and the ecology of programs in various systems. Fascinating stuff, really. I can see why he was hooked when he learned about Tierra; perhaps if I had learned about that really early in my life, I would have chosen a completely different intellectual path.

Anyway, after almost 15 hours of talks in two days, practically living on coffee the whole time, I’m a bit tired. Tomorrow, we get our first rotation assignments!

Yesterday was the first day of orientation for the department, and today, I leave for a departmental retreat, wherein the first-year students get bashed over the head with tons of talks, plenty of poster sessions, free food, drinks, and lots of interaction with the professors, post-docs, and older graduate students.

During the retreat, I’ll also be thinking about this post from “GTD in Academia”, especially the comment about developing a toolkit. (It reminds me very much of the part of Surely You’re Joking, Mr. Feynman, where Feynman talks about his mathematical toolbox and differentiation under the integral sign.) The post focuses on ecology, but of course the advice is more general than that.

Develop a toolkit. You’re going to know how to design experiments and analyze data and think broadly and synthetically about ecology. But you should also develop a toolkit to distinguish yourself from all of the other ecologists who can do those things. Your toolkit might include modeling or null model analyses or genetic techniques or specialized statistics. Just make sure you have one, and make sure everyone knows what it is. MK–make yourself unique and indispensable part of the group.

So I keep wondering, what kind of toolbox can I develop? Even as I try to find a topic that I’m interested in during the retreat, and lab rotation choices, I’ll be thinking hard about that.

All of the professors that I’ve admired have their own little niche that they create with their unique expertise, their esoteric collection of abilities. Howard Berg, for example, is really good at machine-work, and so he was able to hand-make the parts needed for a 3D E. coli chemotaxis tracking microscope, which led to his great theoretical contributions to that field. David Evans is spectacularly good at dissecting molecules down to their basic, synthetically manageable parts, to make clever insights about reaction mechanisms via molecular orbital theory, and visualize asymmetric induction at a very sophisticated level in his head, allowing him to manage the beautiful, almost pedagogic syntheses of really complicated molecules. Martin Nowak is really good at paring away the complexity of a problem to get at the underlying mathematical structure and model, in order to gain very deep, and yet strangely simple and beautiful insight. They’re not expertises in the sense that they’re one-trick ponies; it’s more that they have some sort of edge on the competition that just allows them to break through and do the work better and faster.

I need to find my own toolbox to really succeed and do well. But what? I’m decent at math, programming, physics, chemistry, and biology, but not a big deal on any of it. No subject really scares me, though; I know I can learn more of anything if need be, so I have some help there. But what’ll be my edge? Finding that will be my goal this year. Jack-of-all-trades, master-of-none is not the way to succeed in grad school, I think. Still, until I find that straight-flush, lots of jacks aren’t that bad either…

For the first week of grad school, I have a bit of free time, not having yet joined a lab, so in between walking around and talking to people, I’ve been filling in various gaps in my education. For example, I never really learned statistics — not formally, anyway, my knowledge of genetics and genomics is purely incidental, and my study of computer science and artificial intelligence essentially froze after learning about perceptrons and Hopfield networks in high school.

So, I’ve been reading. A lot. I’ve finally learned the notation and vocab that geneticists commonly use, so that I don’t have to keep googling words as I read papers. It still takes me a half-second to recall stuff (and I still find it strange that the names of many genes are of the mutant phenotype), but it’s definitely sped up my reading of the literature (or at least, the abstract and figure captions).

Statistics is really difficult for me to retain. The problem is, I really like concepts more than details, and I’ve always enjoyed probability theory more than the nitty gritty details of statistics. Unfortunately, everything in science at some point or another needs it, so I learn it begrudgingly, though sometimes learning the details feels like trying to grab a handful of loose sand. Z-test? One-tailed, two-tailed?

Computational biology, on the other hand, I can retain a bit more, perhaps because I’m concentrating on theory and concepts; I definitely can’t remember all the nuances of the various implementations and side-add-ons to the original theory. It’s hard, though, to find good descriptions of basic concepts and theory on the web, rather than implementation. (For BLAST, I eventually resorted to a book; I love O’Reilly, I really do). I’m learning about Bayesian networks from a tutorial I found on “Computational Biology News”, while with singular value decomposition and principal value analysis, I’m learning from a hodge-podge of different sources to try and connect everything together.