Design Automation in Synthetic Biology

Abstract

Design automation refers to a category of software tools for designing systems that work together in a workflow for designing, building, testing, and analyzing systems with a target behavior. In synthetic biology, these tools are called bio-design automation (BDA) tools. In this review, we discuss the BDA tools areas-specify, design, build, test, and learn-and introduce the existing software tools designed to solve problems in these areas. We then detail the functionality of some of these tools and show how they can be used together to create the desired behavior of two types of modern synthetic genetic regulatory networks.

Figure 1.

Specification and design of a genetic AND gate and dual-feedback oscillator. All genetic part symbols are defined by the SBOL Visual standard (Quinn et al. 2015). Parts of the same symbol but different colors have different sequences, whereas parts of different symbols but the same color belong to same genetic logic gate. (A) Cello (Nielsen et al. 2016) takes a Verilog (Thomas and Moorby 1995) specification for an AND gate as input, compiles and colors a gate netlist, and outputs a genetic circuit design (top). GEC (Pedersen and Phillips 2009) takes a GEC specification for a dual-feedback oscillator as input and outputs a genetic circuit design (bottom). The RBS calculator (Salis et al. 2009) outputs a library of RBSs with a range of translation initiation rates (bottom). (B) Eugene (Oberortner et al. 2014) takes Eugene specifications as input and outputs a library of AND gate designs that vary in gene order (top) and a library of oscillator designs that vary in gene order and orientation (bottom). Double Dutch (Roehner et al. 2016a) takes a design of experiments (DOE) matrix as input and outputs a commensurate library of oscillator designs with varying RBSs designed by the RBS calculator (bottom).

Figure 2.

Construction of a complex logic gate and oscillator. (A) Using ApE, one can take a sequence of parts from a design, assign to one or more vectors, and edit any parts or junctions at the DNA sequence level. (B) Raven is used to determine a hierarchical assembly plan for each construct using a specific assembly method. Assembling an AND gate with Gibson requires two stages, four steps, four synthesis steps, and six polymerase chain reaction (PCR) steps, whereas assembling the oscillator constructs with GoldenGate requires two stages, four steps, and 16 PCR steps. Raven generates oligonucleotides to perform these assemblies, but j5 can be used for more refined primer designs for specific cloning steps. The common sequences are highlighted in red. (C) Once a complete assembly plan is determined, PR-PR can be used to create instruction files for liquid-handling robotics or microfluidics platforms to distribute liquids.

Figure 2.

Construction of a complex logic gate and oscillator. (A) Using ApE, one can take a sequence of parts from a design, assign to one or more vectors, and edit any parts or junctions at the DNA sequence level. (B) Raven is used to determine a hierarchical assembly plan for each construct using a specific assembly method. Assembling an AND gate with Gibson requires two stages, four steps, four synthesis steps, and six polymerase chain reaction (PCR) steps, whereas assembling the oscillator constructs with GoldenGate requires two stages, four steps, and 16 PCR steps. Raven generates oligonucleotides to perform these assemblies, but j5 can be used for more refined primer designs for specific cloning steps. The common sequences are highlighted in red. (C) Once a complete assembly plan is determined, PR-PR can be used to create instruction files for liquid-handling robotics or microfluidics platforms to distribute liquids.

Figure 2.

Construction of a complex logic gate and oscillator. (A) Using ApE, one can take a sequence of parts from a design, assign to one or more vectors, and edit any parts or junctions at the DNA sequence level. (B) Raven is used to determine a hierarchical assembly plan for each construct using a specific assembly method. Assembling an AND gate with Gibson requires two stages, four steps, four synthesis steps, and six polymerase chain reaction (PCR) steps, whereas assembling the oscillator constructs with GoldenGate requires two stages, four steps, and 16 PCR steps. Raven generates oligonucleotides to perform these assemblies, but j5 can be used for more refined primer designs for specific cloning steps. The common sequences are highlighted in red. (C) Once a complete assembly plan is determined, PR-PR can be used to create instruction files for liquid-handling robotics or microfluidics platforms to distribute liquids.

Figure 3.

Modeling and simulation results of a genetic AND gate and dual-feedback oscillator. (A) SBML models are created in iBioSim for the AND gate and oscillator. The AND gate’s model is automatically generated by converting the SBOL file produced by Cello to SBML. This model is then extended to include the regulation by the proteins LacI and TetR, the complex formation of LacI and TetR with the small molecules IPTG and aTc, respectively, and the regulation of the florescent protein YFP. The oscillator model is reconstructed using information (e.g., parameter values, rate constants) from the SBML generated by GEC. (B) Simulations are performed using both COPASI and iBioSim. Deterministic simulations are performed using the LSODA ODE solver in COPASI. Stochastic simulations are performed in iBioSim using an implementation of Gillespie’s SSA.

Figure 3.

Modeling and simulation results of a genetic AND gate and dual-feedback oscillator. (A) SBML models are created in iBioSim for the AND gate and oscillator. The AND gate’s model is automatically generated by converting the SBOL file produced by Cello to SBML. This model is then extended to include the regulation by the proteins LacI and TetR, the complex formation of LacI and TetR with the small molecules IPTG and aTc, respectively, and the regulation of the florescent protein YFP. The oscillator model is reconstructed using information (e.g., parameter values, rate constants) from the SBML generated by GEC. (B) Simulations are performed using both COPASI and iBioSim. Deterministic simulations are performed using the LSODA ODE solver in COPASI. Stochastic simulations are performed in iBioSim using an implementation of Gillespie’s SSA.