Well things are wrapping up for my summer research session. I am now in the process of writing up my final project, which essentially includes typing up a summary, the equations, and compiling all of the figures into one place. I also have to do all of my typing in word, which normally is fine for papers, but normally for math-type things I like to use latex. But I downloaded the latest version of word, which has an equation editor and symbols, so hopefully that won’t be too painful.

As far as my project goes though, things are wrapping up pretty nicely. I have a good set of figures from the model and the data analysis that I did on the experimental data, and a replication of a figure from another paper that modeled action selection in the basal ganglia. In addition, I have figures for the output of the model. I also feel very confident in the parameters we are using- all are within the constraints. Very exciting! The next step now, after writing all of this up, is to replicate do some perturbation studies and see if we can replicate experimental data when the populations are ablated. That will be super exciting, especially if it all works out!

This will not be my last blog- I have some things I want to blog about before I turn in my summary. I mainly want to get caught up in my research thoughts, and what’s been going on in terms of where I am in my project, and the bumps in the road that I have experienced.

After doing some thorough reading of the literature, I also did some data analysis from Josh Burk’s data. This data was from a stop signal reaction time task experiment, which tests impulsivity. Normally, the rats have to push a left lever and then the right lever, each one within three seconds. On 20% of the trials, a stop signal will turn on (a light), after the left lever, which signals for the rat to withhold a right lever push response. So, there is data for each of the various latencies, how many times there was a correct or an incorrect response, and if there was a correct start to the trial or not. I was able to plot these latencies as histograms, and had a few bar plots, but there is a great deal of variation in the data that is difficult to replicate in my model. One way we wanted to overcome this issue of experimental variation is through using stochastic processes, but it also may be correct that different rats have different parameter sets.. But then it gets much more complicated because it would be extremely computationally expensive to fit our model to each rat through a parameter study, and it also would definitely not be worth the effort, as our main goal is to use our model to predict things in the experiment, and to figure out how this network of neurons works.

Another issue is what exactly we can predict with the model. There are almost too many options! We are now confident in our model, we can replicate at least a subset of the results, and we have thoroughly analyzed the data from the experiments. But the next step- what would be the most interesting to investigate, is difficult. And it is important to be a prudent researcher- time is precious, and none of it can be wasted if I would like to submit another paper while I am still an undergraduate.

Nerd camp #2 was the Computational Cell Biology course at Cold Spring Harbor Laboratory in Cold Spring Harbor, New York (on Long Island). This lasted a little over two weeks, and it was AMAZING. I met great people, had a great time, but I worked really, really hard as well.

The course started off with a lot of lectures, specifically on modeling dynamical systems. This was sort of covering the basics, and was a good place to start. There were also lectures on modeling chemical reactions, and the general theory behind modeling all types of reactions. The topics generally became more specific as the course went on, and lecturers would discuss their research in say, modeling mitosis, bayesian networks, calcium signaling, modelings synaptic plasticity, etc. These were generally very interesting, although it was quite easy to get lost in the areas that I lacked background information in. Lectures were 9 hours a day, and I spent a hefty portion of my free time working on my own project, which I will discuss in a little bit. Different lecturers came in to speak to us, and it was absolutely amazing to get to listen to these well-established academic heros of mine speak. It was also as amazing to talk to some well-established grad students. I had a lot of great conversations with them, and what they thought of grad school, what they thought of different programs, and so on. I am facing a big transition in my life (from undergraduate to graduate school), and it was really nice to hear what these students had to say, and what advice they had to offer.

Now- for my project. I left the MBI ready to do some serious work on my project- I was ready to continue the intense pace that I was on at OSU. But, I realized that the model has some holes in it that really needed to be cleared up. Some questions that I had don’t have answers yet- which is sort of the point of science- but I did a thorough investigation to answer the ones that had answers. To review, my project concerns building a population activity model of cortico-striatal circuitry underlying impulsive actions. It is thought that there are two main pathways in action selection: the selection pathway and the control pathway (also termed go and no-go). The pathways involve different populations, and it was unclear to me what populations were involved in what ways, and how these pathways eventually converged to give rise to one things: either an action, or not. There are some interesting properties of the populations we are studying as well; for example, the STN is has diffuse synaptic connections, while the Striatum highly converges on one channel (a sort of sub-pathway) of the GP and SNr. So, I really needed to clear all this information and network architecture up if I wanted to build a plausible model. I poured through the literature, and I dug up a ton of information. It was actually a ton of fun- I investigated some new modeling methods, played with Copasi, learned about the architecture, different functions of the basal ganglia (especially it’s role in working memory), and so on. It was really like a mini-independent study on the basal ganglia, but it was exactly what I needed. By the end of the course, I was able to draw out what I thought the architecture was, and I felt really confident about the steps that we were taking with the model. I also began analyzing the experimental data, and fit some free parameters in the model so that it would fit this data.

The next step, which happened in the weeks that followed CSHL, was to provide justification for this model, and link it to the experimental data. This will be the topic of my next blog.