Imitation of Life
Can a computer program reproduce everything that happens inside a living cell?
The computations performed in the transcription module are quite different from those of the metabolic subunit. Instead of linear programming, we have discrete events governed by probabilities.
Transcription of a gene begins when a molecule of the enzyme RNA polymerase binds to a chromosomal site called a promoter. The enzyme then ratchets along the double helix, producing a strand of messenger RNA whose sequence is complementary to that of one DNA strand. When the transcript is complete, the polymerase drops off the double helix and releases the RNA. Each step in this process requires a variety of other molecules—initiation factors, elongation factors, termination factors, energy donors—as well as a supply of nucleotides to be incorporated into the growing RNA strand.
In the WholeCell system, each RNA polymerase molecule is an individual object with four possible states: actively transcribing, bound to a promoter region, bound to DNA elsewhere and unbound. Transitions between the states are random events with probabilities calculated to match the experimentally observed distribution. Also, various promoter sites differ in their affinity for RNA polymerase, so the probability of binding is higher in some places than others.
Because of probabilistic events like these, the WholeCell model has an element of nondeterminism. Every run can be expected to produce somewhat different results, even with the same initial conditions and environment. But of course fluctuations and chance events also have a role in real biology; even perfect clones will not follow exactly the same trajectory through life.
Reading the source code of the transcription module gives some vivid glimpses of the subtleties that a wary modeler must keep in mind. Suppose a roll of the digital dice dictates that a certain RNA polymerase molecule is to bind to a promoter site. What happens if all the promoter sites are already occupied? What happens if two polymerase molecules try to grab the same promoter site at the same time? What if two transcription enzymes collide as they move along the DNA? Nature seems to handle such conflicts without having to think about them, but the modeler has to think of everything.
Collisions between enzymes scuttling along the chromosome are not rare events. Results of the WholeCell simulation suggest they happen about once per second, or perhaps 30,000 times in the course of a full cell cycle. The model is therefore equipped with rules to decide who has the right of way.