Interactive web­ based software for genetic circuit design, full application.

Designing a synthetic circuit from the interconnection of parts or devices can be significantly facilitated by using systematic in silico modeling and design relying on the separation of the design from the actual implementation. In this approach, various designs are first optimised and their properties are assessed using mathematical analysis and model-based computer simulations.

The Idea:

Synthetic biology aspires to design and construct novel biological systems to reprogramme the cell for biomedical or biotechnological purposes. Despite many achievements in the last decade, the sheer complexity of interactions in biological systems has presented insurmountable challenges. The field currently lacks systematic principles for modelling, designing and constructing synthetic circuits that are robust, tunable and scalable. Current design approaches do not take advantage of key links between experimental and theoretical knowledge, and instead rely either on trial and error experimental redesigns of systems (with a limited number of available biological parts and components), or make experimentalists responsible for working with advanced mathematical modeling.

This project aims to develop a rigorous design­modeling web­based tool for synthetic biology using a novel design methodology that has been recently developed in the Control Group (Engineering Department) as part of Peyman Gifani’s PhD project. This design method hides the mathematical complexity of design and modelling and makes it easily understandable for non­technical users. We intend for the method to be accessible and practical for both experimentalists and designers, to help researchers from a variety of backgrounds to tackle the nonlinear dynamics of biological systems. The SynBioFund will help our team to develop essential interactive web­based software that will guide users in designing target circuits, therefore ensuring that the method is freely accessible to the synthetic biology research community.

Using the method, biologists will be able to translate their experiential knowledge into meaningful mathematical representations that will enable them to both predict and control synthetic systems’ behaviours in silico. The software developed with SynBio funding guides the user to gradually translate a qualitative input­output relation to a quantitative representation (nonlinear feedback loops required for the target circuit) via a graphical user interface (GUI). This GUI hide the complexity of underlying dynamical systems theory and makes the required theoretical knowledge accessible to experimentalists. The design tool can be used in two ways: to design new circuits with increased complexity, robustness and tunability and to analyze and explore the capability of currently available circuits to redesign them for better functionality.

The design framework engages nonlinear dynamical systems theory via a novel graphical method in which the user first develops a detailed qualitative description of a target circuit, specifying design criteria or goals for particular cellular input­output relations. The method then guides the user in transforming these design criteria into a specific representational form via a unique graphical palette.

This graphical representation is converted into a qualitative mathematical representation by modeling the system’s trajectories in a phase space. The design framework then specifies building blocks with meaningful biological interpretations, which with the user can translate the qualitative mathematical representation into a mathematical formula representing the circuit structure and its parameters. This formula can then be interpreted as the system’s ordinary differential equation (ODE). This allows the construction of ordinary differential equation­based models, and provides a potential blueprint for implementation of a biological device based on standardised parts. The framework is intended to be used as an iterative process, whereby the qualitative and quantitative representations are gradually refined in order to best fulfil the design criteria.

The proposed project will be a multidisciplinary collaboration between the Engineering Department (Control Group) and the Computer Lab (Rainbow Group), capitalising on the unique intersection between control theory, machine learning, artificial intelligence and synthetic biology. We will consult researchers from the Plant Science and Genetics departments to optimise the user interface design process.

This project provides a valuable contribution to the current field. Designing a synthetic circuit from the interconnection of parts or devices can be significantly facilitated by using systematic in silico modeling  and design relying on the separation of the design from the actual implementation. In this approach, various designs are first optimised and their properties are assessed using mathematical analysis and model­based computer simulations.

This project also can be used as an educational tool for teaching nonlinear dynamical systems theory to student with biological science background. Biologists can understand better the role of nonlinear feedback loops in genetic regulatory systems by designing system using the proposed tool.

Who we are:

Peyman Gifani is a finishing his PhD at the Engineering Department, Control Group. Through his academic path he has enriched his knowledge in multidisciplinary areas including nonlinear dynamical systems modelling, bioengineering, machine learning and software design. The essence of his intellectual and professional life is creative and critical thinking for problem solving. During his PhD in Information Engineering division in the Department of Engineering, he developed a novel design framework based on nonlinear dynamical system theory for synthetic biology. This framework is the basis for developing a web­based design automation tool during this proposal. He has always been fascinated by biomedical engineering, which directed him to develop a new algorithm for brain signal (EEG) processing and raised fund for designing a medical device to monitor depth of anesthesia during surgery before starting his PhD. He is also interested in watercolor painting and his artworks were exhibited at the Queens College Art festival (2013), and in two group exhibitions and a gallery in London (2013; 2014).

Dr. Saeed Aghaee is a post­doc researcher in the Rainbow group at the Computer Laboratory of the University of Cambridge. Saeed Aghaee's research lies at the intersection of human-­computer interaction and software engineering. He has rich expertise in building interactive systems, involving design, prototyping, testing, and back­end and front­end development. He is the main architect of NaturalMash, a complex and innovative natural programming system allowing people to create simple Web mashup applications. He is currently involved in developing a novel natural language programming environment for the Internet of Things.

Implementation:

The goal of the project is to design a web application which automates the design process based on the design framework. Therefore, this project mainly involves front­end and back­end programming.

The web application starts by a detailed description of the desired circuit that specifies target input-output relations. The user interface provides a collection of predefined options for this purpose. The user can start from these examples or define a new set of qualitative description. These target input- outputs relation then should be translated into the language of nonlinear dynamics by finding equivalent local dynamics which can be gradually assembled on a phase space to build a phase portrait. The proposed application specifies a set of 10 local dynamics (sufficient for designing a wide variety of two­dimensional systems), which are graphically represented as nodes on a palette. The user interface enables the user to easily select, arrange, and link appropriate nodes. The application assists users to compile local dynamics into a global dynamic, defining the trajectories of the system in the entire phase space. The design framework provides a set of rules for this compilation. The result is a manifold which is a composite interlinking of these domains. In mathematical terms, individual nodes represent equilibrium points, or the intersection of the system’s nullclines. Therefore, each node can be considered to provide a fragment of each nullcline. Following the framework method, these fragments can be compiled to design complete nullclines that meaningfully correspond to the target system behaviours. This is accomplished in the application by linking the nodes so that the entire phase space is partitioned. The partitioned phase portrait provides a general schematic of the shape of the target system’s nullclines. Finally, the web­based application provides a method for converting this schematic into a meaningful mathematical formula that defines the circuit structure and its parameters. This formula is interpreted as the right hand side of the system’s ordinary differential equation (ODE). The resulting ODE specifies the target system’s feedback structure and a set of parameters corresponding to the desired dynamical behaviours. The proposed application is intended to be used as an iterative process, whereby the qualitative and quantitative representations are gradually refined in order to best fulfil the design criteria. The proposed web­based tool reverses the traditional process of modeling. The traditional methods start by identifying a set of differential equation (based on field specific fundamental laws), and continue by analyzing the result by variety of methods available for parameter estimation and model comparison. Typically, the design of a genetics circuits starts by defining causal relationships between a list of available species (e.g. proteins, mRNA and coding genes). Then it continues by deriving differential equation models based on standard mass­action or Michaelis- Menten/Hill kinetics to describe the time evolution of the species. In the proposed web application, instead of using the traditional method to derive model structure and then explore the whole parameter’s space to hopefully find a set of parameters for the desired dynamical behaviour, the design framework reverses this process and starts from a desired dynamical behaviour and provides a set of model structures and parameter values which can performs that behaviour. For example, if the target dynamical behaviour is a triggerable circuit to produce tunable gene expression pulses or bursting, the framework considers the shape and pattern of responses and provides a set of model structures and parameter sets that can perform this responses; or the design framework provides a set of parameters and model structure in order to redesign an available circuit (e.g. oscillation) to perform a different dynamical behaviour (e.g. bistablity or excitability). In order to accomplish this, the traditional process of phase space analysis is reversed to develop a graphical method for nonlinear dynamical system design. Phase space analysis offers simple yet powerful methods to analyze systems by visualization of the qualitative behavior having the mathematical model of the system. The proposed design method first constructs a phase portrait for a target system considering the qualitative information regarding the input output units, and infers meaningful field­specific mathematical model (differential equations) representing the target processing unit. The result of the design tool will be a set of candidate models and parameters for a desired dynamical behaviour. There might be more than one solution for a dynamical behaviour. The results can be post processed by wide variety of methods available for parameter estimation, as well a few for model comparison to choose the best one considering all the constraints including implementation limitations. The design and modeling will be done by the team. After modeling and design the project will be divided into a core and its sub­projects. The core of the project will be implemented by the team. The budget of the project will be used to hire developers for the subprojects implementation concurrently. The project will have online documentation and user manual. The front­end technologies of the project will be mainly HTML5, CSS3, Javascript and jQuary. The back­end technologies will be java, php, node.js (depending on the design). In this project we combine the aspects of Model­Driven Design (MDD) and User­Centered Design (UCD). We look at the design of proposed synthetic biology design tool as a user­centric model­based development process, in which abstraction and complexity handling through meta- modeling is a key success factor in addressing the heterogeneity of system. In utilizing UCD approach, we will be able to receive early feedback from end­users and apply it on every step of the design process. This is of importance to get closer to the mindset of the end­users as it can help with the design of a more natural tool. We conduct user studies using various methods such as expert review, in­lab and remote usability testing, as well as surveys and interviews.

Benefits and outcomes:

This project aims to develop a freely accessible design­modeling web­based tool for synthetic biology.

The proposed project will be a multidisciplinary collaboration between the Engineering Department (Control Group) and the Computer Lab (Rainbow Group), capitalising on the unique intersection between control theory, machine learning, artificial intelligence and synthetic biology. The proposed design­analysis­construction tool is applicable to any field tackling nonlinear dynamical systems, with immediate application in synthetic biology. This project also can be used as an educational tool for teaching nonlinear dynamical system theory in both fields of engineering and biology. It can be extended in future to include circuit integrations and parts recommendation. Therefore, we believe that the project fits the remit of the Synthetic Biology SRI and the judging criteria.

Budget:

The budget of the project will be spend on ­­ programming cost (not for the team): 3500£ ­­ On­demand cloud computing facilities and web hosting: 500£ The primary applicant of the project is going to start a postdoc in Cambridge which provides additional support for the future extension of the project.