Why I’ve Switched to Biophysics

Along with moving to a new institution for my Ph.D., I have decided to switch fields, moving from hard condensed matter/ nanoscience to soft condensed matter/biological physics. This decision was totally unplanned - even when applying to and visiting graduate schools, I thought I wanted to do some variant of what I did as an undergrad. The thing is, as I visited more and more schools and learned about different people’s work, I found that what really captured my interest and got me excited was the biological side of things. Things tend to be messier and less well understood, but this just means that there’s more to learn and quantify, and I think physicists are well-positioned to bring something new and useful to the table.

Two months later, I’m spending all my time either reading about, talking to people about, or working on biology/biophysics experiments. It’s a lot of fun. Research-wise, I’m currently interested in the mechanics of cells and aggregates of cells (like biofilms); broadly speaking, I want to learn more about how they interact with each other and with their environment.

While in one sense my interest in soft condensed matter and biological physics has been steadily increasing over the past few years due to some really great classes I took as an undergrad, there are three key experiments or ideas that I’ve come across in the past two or three months that solidified my falling in love with this field.

1. Tensegrity and the structure of biological systems
Tensegrity is a mechanical design principle pioneered by Kenneth Snelson and Buckminster Fuller in the 1960’s, in which structures are designed such that the competition between forces - tension versus compression - throughout has a self-stabilizing effect. (A well-known example of this is the geodesic dome.) Among others, Don Ingber has spent a lot of time exploring the application of this idea to the structure of cells. Basically, the idea is that the cytoskeleton of the cell is composed of a network of interconnected units - the microfilaments, microtubules, and intermediate filaments - under tension and compression; that is, it is structured according to the principle of tensegrity. Many groups have explored this idea since it was first proposed, and other theories exist for understanding cellular structure; indeed, many groups, including the one I’m in, currently spend a lot of time trying to better understand the structure and physical properties of cells. (You can read more about this here.) I just thought the idea was so darn cool when I first came across it in this very nice Scientific American article written over a decade ago (look at the cell on page 54!). What’s more, this idea could be used to understand the structure of other assemblies at the micro- or nano-scale, such as buckyballs or nanotubes (e.g. see the chapter by Yakobson on “Carbon Nanotubes: Supramolecular Machines” in the Dekker Encyclopedia of Nanoscience and Nanotechnology), actuated nanocolumns, and even…

2. Viruses from a materials perspective
Yep - reading about tensegrity led to me to Caspar and Klug’s classical work in the 1960’s, in which they attempted to understand the structure of ’spherical’ viral capsids within a tensegrity-inspired framework. Since then, a number of physicists and engineers have spent a good deal of time trying to understand the structure of viral capsids. One framework in particular, developed by David Nelson and co-workers, really appeals to me: I think it’s an elegant combination of ideas from crystallography and continuum mechanics (what they call “spherical crystallography”). Basically, the idea is that if you try to pack a number of particles - be they beads, or the protein subunits of a viral capsid - on the surface of a sphere, the resulting assembly necessarily possesses crystallographic defects resulting from geometrical frustration. I wrote a small review of viral structure and mechanics focused on this work for a nanomechanics class not too long ago, which you can read here, if you want to explore this further. And this is just the tip of the iceberg - people are doing all sorts of crazy things with viruses: playing tug of war with them, watching them spit out their DNA, poking on them, shocking them, and filling them with various cargoes, among other things. Pretty cool stuff.

3. Hitting worms with laser pulses
Again, this is a very broad field in which a lot of great work has been (and continues to be) done. I don’t know enough about it. What first got me excited about biological neural networks and c. elegans was learning about this experiment by Mehmet Fatih Yanik. Basically, Yanik et al. used femtosecond laser pulses to cut single axons in c. elegans worms, observed the resulting phenotypic effects, and watched them grow back within 24 hours. This is pretty neat - after all, being able to perturb these affords researchers quite a lot of control, and could be used to study nerve regeneration processes one axon at a time, among other things. c. elegans is quite the model system, and I’m sure there’s a lot of other cool work going on trying to understand various processes and mechanisms in these worms. For example, in addition to Yanik’s work, my very little reading of research in this field has exposed me to some very interesting papers from Richard Morimoto’s, Ikue Mori’s, Aravi Samuel’s and William Bialek/William Ryu’s groups, to name a few off the top of my head. I still need to learn more about this field, particularly of what the biologists are doing - but again, this femtosecond laser stuff really got my attention when I first came across it.

Cool Papers 1: General

I’ve come across a number of pretty cool papers in the past few months. Some of them deal with particular phenomena (stay tuned for possible upcoming posts on molecules at surfaces, biomimetics, phononics, crystallization, nanoparticles, wetting phenomena, computational physics, etc. etc. - at some point), and so are probably better off getting their own blog posts. Here are a few papers that didn’t fall into specific categories…

1. Frictional Anisotropy on a Quasicrystal Surface
Along with ~10 other things, a subject that I’ve recently become interested in is nanoscale mechanics, broadly defined. A key experimental tool in this field is the use of local probes to push or pull on things controllably. Miquel Salmeron’s STM group at Berkeley does work on this and related subjects, and I finally got around to reading this paper of theirs from a few years back.

The idea is conceptually very simple: while friction unsurprisingly depends on commensurability (that is, if two surfaces in contact are structurally ‘complementary’, they will ‘lock in’ to each other and hence have high friction between them - an idea that apparently dates back to da Vinci), trying to think about friction using just this notion is unrealistic. For starters, most contacting surfaces are probably incommensurate, and other factors - such as periodicity(?) - contribute, as well.

This paper nicely singles out the role of periodicity by looking at different directions along Al-Ni-Co quasicrystal surfaces using STM (to image the surface and hence distinguish the periodic and aperiodic directions of atom ordering) and AFM (to measure the probe tip-surface friction along these directions) in ultra-high vacuum. The AFM friction data can be modeled using a classical model relevant to the experimental situation (the Derjaguin-Muller-Toporov or DMT model, which I need to learn more about), enabling key parameters to be derived from the measurements.

In particular, the authors find a larger friction force (8x) along the periodic direction than along the aperiodic direction. Unsurprisingly, they ascribe this to differences in energy dissipation via electron or phonon excitation+propagation along the different directions, although it is unclear to what extent each kind of excitation plays a role. Perhaps similar local-probe measurements of a different kind (e.g. ones sensitive to electrical versus mechanical properties) might be useful… At the end of the day, I like this paper because it is an elegant example of using a unique microstructure, in which just one variable (here periodicity) changes in ways that are well understood, to study something interesting as a function of just that variable.

2. Liquid Crystals and the Origins of Life
Noel Clark gave a great talk about this work here at Penn not too long ago. I won’t write too much about this since Randy has a nice description of it over at the condmat journal club.

Here’s the executive summary: according to extensions of Onsager’s rigid-rod model for the formation of liquid crystal phases, individual molecules must be sufficiently anisotropic (i.e. the aspect ratio has to be above a certain minimum) to form a liquid crystal (LC). Surprisingly, the authors of this paper observed LC phases consisting of single-stranded (ss) DNA molecules too short to satisfy this criterion. Optical and x-ray measurements indicate that this results from end-to-end stacking of duplexes of complementary short ss-DNA molecules (known as ‘living polymerization’) into larger rods that satisfy the Onsager criterion, even at low temperatures (in concentrated phases of duplexes separated from the isotropic phase of unpaired ss-DNA molecules).

This autocatalytic behavior is like positive feedback, in a sense, and is why this work is so interesting from a biological point of view: it provides a mechanism by which the right molecules can be ’selected’ out from a ’soup’, and ‘evolve’ into larger ones as part of an RNA world. It’s an interesting idea - definitely one that’s gotten a lot of press, it seems - and while this work doesn’t provide much hard evidence for it, I’ll be interested to see what it stimulates.

3. Suprafroth!
This is a very interesting paper out recently on the arxiv, I think to be published in Nature Physics. While I don’t understand all the details, I like this particularly because it’s a nice combination of ideas from soft- and hard-condensed matter physics, like electronic liquid crystals.

The authors used magneto-optical imaging, which I need to learn more about, to image the flux pattern of superconducting lead (a type-I superconductor). Turns out that the magnetic field on the edge of a disc-shaped sample of lead is larger than the actual applied field, and for large enough magnetic field some flux can penetrate the sample. This leads to a phase intermediate between the normal and superconducting phases, possessing a froth-like magnetic structure - specifically, the froth cell boundaries are superconducting, while the interiors are normal metal. This shows up very clearly in the magneto-optical images (see figures in the paper).

The nice thing is that, unlike ‘conventional’ froths, mass-transport processes like drying or drainage are not present here (as the authors point out, “this superconducting froth involves only electrons”). This means that the froth structure can be tuned reversibly using the applied magnetic field or temperature, and the nice magneto-optical images allow for quantitative analysis of the froth structure as a function of just these parameters.

This is philosophically similar (loosely speaking) to paper #1 - the friction measurements of quasicrystals: again, it is a very nice example of using a unique microstructure (here, a froth structure that doesn’t suffer from irreversible processes, and can be controlled by magnetic field or temperature) to study something interesting (here, the structure and dynamics of froths) as a function of just the variables that you can control.

4. Universality in Conference Registration
This is a cute correspondence recently sent to Nature Physics describing an intriguing social application of statistical mechanics.

The authors used registration data from two physics conferences (# of registrants as a function of time to the deadline), saw that they matched up remarkably well (after rescaling), and came up with a simple model to capture the observed phenomenon in which the ‘pressure’ felt by potential attendees to register varies inversely with respect to the time to the deadline. Also, incorporating a Boltzmann-like factor (instead of uniform probability to register over the period of time) leads to a prediction that agrees well with # of payments as a function of time to the deadline data.

Of course, there are a number of assumptions and fitting parameters floating around here, and I’m not entirely sure this work will change the world of physics, but I always find things like this fun.

Graduate School

It’s graduate admissions season, apparently. So far I’ve heard back from six institutions: four acceptances (including a number of my top choices) with fellowships at two of them, one we’ll-fly-you-out ‘interview’, and one phone interview that went rather well… so at the end of the day, figuring out where I want to end up may be nontrivial.

As I noted in a previous comment, my algorithm for deciding where to apply to was pretty simple: I spent a good deal of time soul-searching and deciding (roughly) what I want to do for the next five-ish years, made a list of all the people I thought would make good research advisors to that end, and applied to the n departments that contained the maximal number of the people on my list (where n was determined by how many applications I was willing to fill out; turned out to be thirteen in total - I’m not superstitious).

My algorithm for deciding where to go for grad school will probably be a variant of the algorithm I used to pick my undergrad institution, and again is not too complicated (Sean Carroll has a very nice post on this subject, btw, as does okham as I just discovered):

1. Make a list of all the factors that I care about: for example, number/quality of advisors who I’m interested in (based on various factors like personal interactions, reputation, publication record, how they place people…), how excited I am by current research efforts, intellectual environment, potential for interesting collaborations, other students, location, quality of the department/school/life, bureaucratic requirements, funding, etc.

2. Weight individual factors accordingly: pretty self-explanatory, although this requires a lot of thought.

3. Visit all the places I’m seriously considering/find out as much about them as possible: this is the data collection stage, so that I have a good idea of how various places shape up in terms of the factors I listed. I’m pretty much traveling every weekend from next Friday to the end of March, with a few days in between for the APS March meeting. That’ll be fun.

4. Assign data values corresponding to each factor for each department: i.e. the results of step #3. These data values will obviously have error bars to reflect the subjectivity inherent to the data collection process, but the inverse relationship between error bar size and time spent on step #3 should enable a single-valued result.

5. Plug and chug: go wherever above algorithm says to go. Hey, it worked pretty well for my undergrad.

Speaking of which, I feel compelled to plug Penn. If anyone reading this happens to be a senior who got into Penn (terminology: I refer to UPenn, not Penn State) for something physics / materials science / nanoscience-related, I strongly urge you to think about coming here for grad school. There’s a lot of very exciting work going on here, and a lot of great people to work with - fantastic intellectual environment (I would say Penn does pretty highly on all the factors I mentioned above).

Whew.

I realize that I pretty much dropped off the face of the earth for a while - the result of a crazy semester filled with endless problem sets, writing papers (both research and for classes), working on experiments/having equipment break down on me, the GREs, grad school apps, and random personal issues. Thankfully things are far less stressful nowadays what with the holidays + new semester and all…

One of the highlights was going to Boston to talk about some of my recent work at the Materials Research Society fall meeting. It was great - I had the chance to meet some very interesting people and go to quite a few exciting talks, both of relevance to my research or of personal interest. I might post about some of the ideas/papers I’ve been thinking about since then, some partly stimulated by some of the MRS talks or talks we’ve had here at Penn, some partly stimulated by my recent experiments.

Random thought: a service that I would love to have (I’ve found myself wishing for this kind of thing both when digging through the literature on a particular subject) is an online ’science network visualizer’ application. Basically an online service (maybe provided by ISI, Google scholar or something of that sort?) that would take a given researcher’s name - or perhaps an individual paper citation - as an input, and generate a ‘map’ of what people/groups of people/other papers cite them, and with what frequency. I know facebook has a few things like this (e.g. the many eyes friend network visualizer), and the awesome people at information esthetics created this map of science (of which I have a poster) using a similar algorithm. This kind of service would be invaluable for figuring out who is paying attention to a particular kind of research (and what else they pay attention to), what ‘cliques’ exist in various scientific communities, what the big results are/who the big players are, what connections are just waiting to happen… actually now that I think of it, this kind of thing probably already exists, hopefully in a more general framework that can be used to visualize other kinds of communities as well. I am an avid rss subscriber of visual complexity, a great site by Manuel Lima that documents various kinds of network visualization, but I don’t think I’ve ever come across something that fits this description.

In other news, I’ve started to write my senior/master’s thesis and have had to relearn . At the beginning the learning curve was incredibly frustrating, but things are finally starting to fall into place. After much experimenting, I’ve decided to go with BibDesk for managing citations, as suggested by Andrew Dawes (of The Daily Photon, a valuable new addition to the blogroll). It’s a wonderful program, and now that I’m doing more on the front it makes a lot of sense. Andrew has a post on some mac apps that he uses - similar to my previous post on the subject, and presumably with more to come.

Speaking of which, another new addition to the blogroll is confused at a higher level, an interesting “professional journal” by a physicist in Minnesota. He has a few posts that I thought were particularly interesting, for example documenting various aspects of the peer-review process (like ref comments) or collaborations that he’s working on. Also, he’s a Tufte fan (see my post from ages ago).

Keeping Track of Publications

This is frustrating. While I’m usually pretty good with keeping up with the latest publications and preprints (via journals’ rss feeds), I like to follow the work of particular groups as well. I do this by going to the group’s webpage (which never works - hardly anyone seems to keep an accurate publication list) or searching on ISI Web of Science (which is far too time-consuming). ISI appears to have an option where they email you updates to your latest saved search, but I can’t figure out how to make it do what I want.

Ideally, there would be an online service (which covered all physics, chemistry, materials science etc. journals) which would allow me to input an unlimited number of author names, and would email me (or provide an rss feed) with updates to their citation record as they occurred. For free. Maybe such a thing already exists, but I haven’t found it.

Update: Via an overly complicated combination of stringing together search terms and saving histories, I managed to figure out a solution on ISI (I think - I’ll have to wait until the first email alert to be certain). As psi*psi points out, Yahoo Pipes may be a viable option, but I haven’t had the time to play around with it to figure out what it can and can’t do.