Design of Languages for Systems and Synthetic Biology
Context
We propose to encompass context-dependent genomic information and mathematical models in a logical and structured fashion through the use of attribute grammars.
Attribute Grammars (AG) provide means to compute a mathematical model (possibly encoding phenotypic traits) from a genetic construct. AGs are a framework for biological Domain Specifi c Languages (DSLs) that, unlike current formalisms to associate a genotype to a phenotype (databases, ontologies), can confer predictive powers. Hence, the language’s genetic constructs can be studied in silico, an approach which is increasingly essential as synthetic biology becomes more complex, but which also makes it possible to predict phenotypes of mutants based on their genome sequences.
Methods
DNA compilers based on such AGs can analyze a genetic construct and output the corresponding mathematical model according to the language semantics. We propose a methodology for the development of DSLs to design and analyze genetic constructs. Because development of those languages is dependent on current biological understanding of the modeled mechanisms, it can be incrementally refi ned for the target biological application.
We implemented the work ow to design a DSL, generate the DNA compiler, use the language to design genetic constructs, and analyze constructs in silico by running time course simulations of the produced mathematical model in GenoCAD — a point-and-click CAD tool for synthetic biology relying on formal language theory. Using GenoCAD, a Context-Free Grammar (CFG) can be designed using a drag-and-drop interface; for AGs, attributes can also be specifi ed, and a library of preset functions can be used to make declarations for the semantic actions, or to set attribute values. Resulting biological languages are stored in a relational database. We then propose to generate the semantic DNA compilers on-the-fly from user speci fications to analyze their designs.
Results
We illustrated our work ow with three Synthetic Biology Markup Language (SBML) grammars. In the fi rst grammar, rate laws are modeled by Wilson Cowan equations. This language allows users to design and analyze gene regulatory networks with up to three interacting proteins inserted into plasmids, and we reproduced both the genetic toggle switches and the repressilator with the same library of genetic parts. In the second grammar, we showed we can easily change the rate laws and the granularity of design by generating Mass Action equations instead. We then design an application-speci c language to design AND gates and layer them to make a 3-input AND gate. We also showed that other modeling approaches could be used in the DSL. We wrote an attribute grammar that outputs a boolean logic model of the gene regulatory network in GinML format and reproduced the simpli ed cI/cro system of the lambda bacteriophage. Finally, we point out that by re-using the work from system biologists, these DSLs can also model natural genomes.
Tools
Editor (+compiler generation)
Upcoming. Please, refer to GenoCAD and the ‘More information?’ section meanwhile.
APIs
More information?
- Read my full thesis chapter: chap6thesis
- Read the additional pages ‘Systems Biology: Biological Languages to Design Natural Genomes with the Cell Cycle Example’ cellcycleAG
- See a seminar I gave
- The grammar files are available at http://figshare.com/authors/Laura_Adam/388642