This paper deals with a model of cellular growth called "Epigenetic Tracking", whose key features are: i) distinction bewteen "normal" and "driver" cells; ii) presence in driver cells of an epigenetic memory, that holds the position of the cell in the driver cell lineage tree and represents the source of differentiation during development. In the first part of the paper the model is proved able to generate arbitrary target shapes of unmatched size and variety by means of evo-devo techniques, thus being validated as a model of embryogenesis and cellular differentiation. In the second part of the paper it is shown how the model can produce artificial counterparts for some key aspects of multicellular biology, such as junk DNA, ageing and carcinogenesis. If individually each of these topics has been the subject of intense investigation and modelling effort, to our knowledge no single model or theory seeking to cover all of them under a unified framework has been put forward as yet: this work contains such a theory, which makes Epigenetic Tracking a potential basis for a project of Artificial Biology.
Deep Dive into Epigenetic Tracking: Towards a Project for an Artificial Biology.
This paper deals with a model of cellular growth called “Epigenetic Tracking”, whose key features are: i) distinction bewteen “normal” and “driver” cells; ii) presence in driver cells of an epigenetic memory, that holds the position of the cell in the driver cell lineage tree and represents the source of differentiation during development. In the first part of the paper the model is proved able to generate arbitrary target shapes of unmatched size and variety by means of evo-devo techniques, thus being validated as a model of embryogenesis and cellular differentiation. In the second part of the paper it is shown how the model can produce artificial counterparts for some key aspects of multicellular biology, such as junk DNA, ageing and carcinogenesis. If individually each of these topics has been the subject of intense investigation and modelling effort, to our knowledge no single model or theory seeking to cover all of them under a unified framework has been put forward as yet: this w
This paper is concerned with a model of cellular growth called "epigenetic tracking" (described in (Fontana, 2008)), that belongs to the field or Artificial Embryology or Computational Development. The model is tested by its ability to generate arbitrary target shapes by means of evo-devo techniques, task which is taken as a measure of the model goodness and at which it appears to be quite successful. Subsequently, the implications of the model are explored, in relation to some key aspects of cell biology: embryogenesis, junk DNA, ageing and carcinogenesis: the model is shown able to produce artificial counterparts of each of these aspects (albeit with a reduced level of complexity). The paper is divided into two parts. The first part reviews the previous work in the field of artificial embryology (section 2.1), describes the model of cellular growth (section 2.2) and reports the results of the experiments performed (section 2.3). The second part shows how the model is able to generate artificial counterparts for each of the following key aspects of cell biology: embryogenesis (section 3), junk DNA (section 4), ageing (section 5) and carcinogenesis (section 6). Each of these sections opens with a description of the experimental evidence relevant to the aspect of biology considered, reviews the main existing models and theories and finally describes the artificial counterpart produced by epigenetic tracking. Scattered among more aspects, the topic of stem cells is also discussed. Finally, section 7 draws conclusions and outlines future research directions.
The previous work in the field of Artificial Embryology (see (Kumar and Bentley, 2003;Stanley and Miikkulainen, 2003) for a comprehensive review) can be divided into two broad categories: the grammatical approach and the cell chemistry approach. The grammatical approach, originated by Lindenmayer (Lindenmayer, 1968), evolves sets of rules in the form of grammatical rewrite systems; the grammar can be context-free or context-sensitive and can utilise parameters; variations on this theme include using instruction trees or directed graphs in place of actual grammars. L-systems were employed as a means of describing the complex fractal patterns observed in nature and particularly the architecture of plants. The cell chemistry approach draws inspiration from the early work of Turing (Turing, 1952), who introduced a mathematical model of diffusion and reaction within a physical substrate. This approach attempts to mimic more closely how physical structures emerge in biology; cells are arranged in a physical space where simulated proteins can be sent as signals from one cell to another, as in nature.
Within the grammatical approach, Sims (Sims, 1994) used directed graphs to evolve the body morphologies and neural networks of artificial creatures in a simulated 3D physical world; in these graphs, a node represents a body part and an edge specifies how body parts are connected. Using a domain similar to Sims’, Hornby and Pollack (Hornby and Pollack, 2002) applied L-systems to the simultaneous evolution of the body morphologies and neural networks of artificial creatures in a simulated 3D physical environment. Cangelosi, Nolfi and Parisi (Cangelosi et al., 1994) devised a model of neural development which includes cell division and cell migration in addition to axonal growth and branching; the development process shows successive phases of functional differentiation and specialisation. Gruau’s Cellular Encoding (Gruau et al., 1996) uses grammar trees to encode steps in the development of a neural network starting from a single ancestor cell; the grammar tree contains developmental instructions at each node.
Within the cell chemistry approach, Random Boolean Networks (RBN’s) were originally developed by Kaufmann as a model of genetic regulatory networks (Kauffman, 1969); in the context of the development of multicellular organisms, the attractors of RBN’s are interpreted as the different “cell types” of the organism. De Garis (De Garis, 1999) developed a model for evolving shapes in 2D reproductive cellular automata; the model was successful in evolving convex shapes but non-convex shapes (e.g. the L-shape) presented a problem. Bongard and Pfeifer (Bongard and Pfeifer, 2001) proposed a minimal model of ontogenetic development to evolve both the morphology and neural control of agents that perform a block-pushing task in a physically-realistic, virtual environment. Inspired by the cell adhesion process, Hogeweg (Hogeweg, 2003) developed a model to simulate morphogenetic processes such as cell migration or engulfing, achieving to evolve complex artificial organisms. Miller and Banzhaf (Miller and Banzhaf, 2003) developed artificial organisms (the french flag) based on a method called Cartesian Genetic Programming, which evolves a developmental program inside cells.
In our model the phenotype of the organism is represented as a 2-dimensional array of square-shaped c
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
