Introduction

If we would know how the transition between the prebiotic world and the advent of the first living cells could have happen our attempt to create artificial cells de novo would be much simpler. Today after more then thirty years of high-performance computing not even computer-simulations can conceivably demonstrate such a transition. Though we now have a considerable knowledge of the molecular details of life as it is today, experimental studies [35, 28], were not able to create life from scratch in a laboratory. Apart from the project PACE most of these initiatives require the existence of enzymes. Minimal cells [19] or synthetic biology (c.f. Synthetic Biology) carry more than 3 billion years of evolutionary history which is much to complex to really understand this transition. Worse, theoretical studies had to make a lot of high-level assumptions which could not help the experimentalists to make better decisive experiments.

The first serious scientific modeling of self-reproducing systems has been devised with replicating numbers [4] and tried to deal with the problem of an automaton being able to self-replicate [53]. Life as a game marked the simplest model in using cellular automaton platforms concerning artificial self-reproducing systems [7]. In the seventies, Holland, the inventor of Genetic Algorithms and Classifier Systems [22, 24], created α-universe [23] to model the genetic replication apparatus in a one-dimensional simple string processing system. McMullin later could show that side-reactions destroy the self-reproducing capabilities of entities in this model-system [34].

The first observed computer-viruses sparked research in evolving software which became prominent with the game Corewars [10], a collection of programs working in parallel on a one-dimensional circular string as was the case with Holland's α-universe. Coreworld [40] and Tierra [42] followed and could at least show phenotypic behaviors found in the Cambrian Explosion.

From a biological point of view, experimental studies in the sixties and seventies [45] led to the development of the quasi-species theory [13], showing that replication errors limit the amount of information which can be stabilized over generations (error threshold). Furthermore, Artificial Chemistry took a more operator-based view of chemistry on the self-replicating computer entities [32, [14], see [11] for a review.

All these experimental and theoretical studies on replication and evolution culminated in the endeavor to really build artificial cells, see ProtoCells http://protocell.org for an overview on some of these projects. Especially the PACE project asks the question how computer-science can contribute to the information processing in primitive artificial cellular organisms. One avenue in getting artificial cells realized is described in [41, 39]. The central question in artificial cell research is, how hereditary information in these cells can be conserved over time and can make an essential contribution to the survival and robustness of these primitive entities.

The work presented here will show three results: a set of universal micro-controllers, though simple, showing the software executed by these controllers being able to stably sustain a software-replication system. The second and most important result concerns the emergence of these replicator-programs without seeding the system other than with random-numbers. The third result is more of an indirect nature, namely, the simplification of the system - which only was possible using reasonable physical assumptions (see embodiment literature, e.g. [38, 44]). Especially the dependency of the dynamic properties from the used simulation-parameters give important insights into the molecular environment. 1


Footnotes:

1 Concerning the nomenclature: technico termini from computer-science and biochemistry are used throughout, e.g. a genome here means a piece of software if not otherwise noted. This mixing should make clear the similarity of concepts and ease the bridging between the two fundamentally different disciplines. In any ambiguous case the connotation meant is specifically mentioned.