An even *HUGER* (14.5Mb) mpeg file showing
the simulator in action
Tubulin is protein dimer - a small double molecule made up of two very similar 'tubulin monomers' (alpha tubulin and beta tubulin), with slightly different properties. In the cell (and in a pure solution, under the right conditions) these monomers will bind together with each other to form large (compared to the size of tubulin) structures known as 'microtubules'. When they form these structures, the beta tubulin in the dimers eventually undergoes a chemical change to become a less stable (less 'sticky') form, which tends to remain stable only so long as it is embedded in the larger structure of the microtubule.
Microtubules are long, hollow tubes, that are arranged out of a large number of alligned tubulin dimers. The tubules are usually 13 dimers in circumference (although they can vary between 12 to 16 dimers depending on environment, especially ion concentrations), and can be many thousands of dimers in length. The actual details of this self - assembly are not entirely clear. There is some argument about whether the tubulin dimers extend in a spiral fashion, or lengthwise, or some combination of both. In addition, the method by which microtubules break down is not entirely clear either - they appear to 'fray' at the ends, and in a sudden and very rapid process disintegrate.
After some suggestions and help from my supervisors, I set out to see whether it was possible to model this behaviour, by simply starting with a collection of loose dimers, with some simple behavioural rules (and, at least initially, some simplfying assumptions).
At the moment the various tubulin binding parameters are set in code - I'm hoping to produce a version soon with a nifty GUI that will allow the various 'stickiness' factors to be adjusted when the program is run (possibly from the web, but no promises!) This will make it easier to see what combinations produce what behaviour. At the moment, for example, it seems that spiral growth is far more stable than longitudinal growth - but this may simply be a matter of choosing the right variables. Or it may not, in which case I may have shown something useful...
Here are some of the early results from the program:
I'm trying to set things up so that the 'tubulin' program can be run over
the web, via a cgi script. It's still early days, but feel free to
try it!
Back to my home page
- A time series
(~130k)
- A large run of the
model (10,000 dimers) (~160k) 
- A large ray traced
picture, (~160k)