A semantic expert system for the evolutionary design of synthetic gene networks
2014
This work tries to cope with the standardization issue by the adoption of model exchange standards like CellML, BioBrick standard biological parts and standard signal carriers for modeling purpose. The BioBricks are easily assemblable [1] standard DNA sequences coding for well-defined structures and functions and represent an effort to introduce the engineering principles of abstraction and standardization in synthetic biology. Web applications as GenoCAD [2] are available and implements an algorithm of syntax check of the circuits designed [3], while some other tools for automatic design and optimization of genetic circuits have appeared [4] and are also specific for BioBrick systems [5]. Our generated models are made of Standard Virtual Parts modular components. Model complexity includes more interaction dynamics than previous works. The inherent software complexity has been handled by a rational use of ontologies and rule engine. The database of parts and interactions is automatically created from publicly available whole system models. We implemented a genetic algorithm searching the space of possible genetic circuits for an optimal circuit meeting user defined input-output dynamics. The tools performing structural optimization usually use stochastic strategies, while those optimizing the parameters or matching the components for a given structure can take advantage of both stochastic and deterministic strategies. In most cases it is however necessary human intervention, for example to set the value of certain kinetic parameters. To our best knowledge no tool exists which does not show a couple of these limitations, then our tool is the only capable of using a library of parts, dynamically generated from other system models available from public databases [6]. The tool automatically infers the chemical and genetic interactions occurring between entities of the repository models and applies them in the target model if opportune. The repository models have to be modeled by a specific CellML standard, the Standard Virtual Parts (SVP) [7] formalism and the components have to be annotated with OWL for unique identifiers. The output is a sequence of readily composable biological components, deposited in the registry of parts , and a complete CellML kinetic model of the system. Accordingly, a model can be generated and simulated from a sequence of BioBrick, without any human intervention. Actual tools present a moderated degree of accuracy in the prediction of the behavior, principally due to the lack of consideration of many cellular factors. Despite the advances in molecular construction, modeling and fine-tuning the behavior of synthetic circuits remains extremely challenging [8]. We tried to cope with this issue of scalability by means of ontologies coupled with a rule engine [9]. Model complexity includes more interaction dynamics than previous works, including gene regulation, interaction between small molecules and proteins but also protein-protein and post-transcriptional regulation. The domain was described by using Ontology Web Language (OWL) ontologies in conjunction with CellML [10], while complex logic was added by Jess rules [11]. The system has been successfully tested on a single test case and looks towards the creation of a web platform [12].
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
9
References
0
Citations
NaN
KQI