For this years Halloween I presented the mathematical biology seminar at the Centre for Mathematical Biology. Here is the title and the abstract…
Cycles in finite populations: a reproducible seminar in three acts
Many natural populations exhibit cyclic fluctuations. Explaining the underlying mechanisms of such cycles is a central problem in ecology and has preoccupied population ecologists ever since Elton’s classical work in 1924. Over the years, a wide range of mathematical models have been explored in an attempt to gain understanding of the conditions giving rise to or inhibiting population cycles. Many of these models, however, rely on the assumption that population sizes are infinite, and hence implicitly assume that the effects of demographic stochasticity are negligible.
Here I will show how demographic stochasticity can give rise to regular and persistent population cycles, so-called quasi-cycles, in simple finite consumer-resource models that are deterministically stable. The existence of such quasi-cycles expand the scope of population cycles caused by ecological interactions, thereby complicating the conclusive interpretation of such patterns. I will discuss how quasi-cycles dovetail with existing theory and will also illustrate the feasibility of accurately identifying such cycles by systematically applying a series of simple analyses to simulated data and data from natural populations.
I will be using this presentation to illustrate how reproducible computational science can be practiced.
Regarding the final statement about reproducible computational science… In the name of transparency and executability (if you look at the slides you will understand why this is important) I am making all the information necessary to reproduce this work available here, i.e. slides, source for slides (Sweave code), simulation code (in C), almost all of the data, figures, etc.
While I had (and still have) the intentions of making all the data available the sheer size of the data in this case (60000 files totalling around 18GB) is problematic in terms of sharing. At this point I am simply not able to post this amount of data online. This is clearly going to be an issues in similar future projects so I need to find a solution. The good news is that you are able to recreate the data by running the simulation code (but it takes a while). If you really want to original data I would be happy to provide it but you will probably have to mail me (mail like in envelope and stamp mail) a sizeable memory stick. Please contact me first to make arrangements. Of course, I would also be interested in hearing any possible alternative solutions to this conundrum.
A few words about reproducing this research. The slides (and hence all the research) is done using Sweave with Beamer so you will need to have R (2.13.1) and LaTeX installed. The code for the stochastic simulations is in C (based on the code here) so you’ll need to have an appropriate compiler (i.e. gcc), unless you are running OS X Lion in which case you might be able to use the included binaries.
This is from the “Mario’s Entangled Bank” blog ( http://pineda-krch.com ) of Mario Pineda-Krch, a theoretical biologist at the University of Alberta.