- Open Access
TinkerCell: modular CAD tool for synthetic biology
© Chandran et al; licensee BioMed Central Ltd. 2009
Received: 04 July 2009
Accepted: 29 October 2009
Published: 29 October 2009
Synthetic biology brings together concepts and techniques from engineering and biology. In this field, computer-aided design (CAD) is necessary in order to bridge the gap between computational modeling and biological data. Using a CAD application, it would be possible to construct models using available biological "parts" and directly generate the DNA sequence that represents the model, thus increasing the efficiency of design and construction of synthetic networks.
An application named TinkerCell has been developed in order to serve as a CAD tool for synthetic biology. TinkerCell is a visual modeling tool that supports a hierarchy of biological parts. Each part in this hierarchy consists of a set of attributes that define the part, such as sequence or rate constants. Models that are constructed using these parts can be analyzed using various third-party C and Python programs that are hosted by TinkerCell via an extensive C and Python application programming interface (API). TinkerCell supports the notion of a module, which are networks with interfaces. Such modules can be connected to each other, forming larger modular networks. TinkerCell is a free and open-source project under the Berkeley Software Distribution license. Downloads, documentation, and tutorials are available at http://www.tinkercell.com.
An ideal CAD application for engineering biological systems would provide features such as: building and simulating networks, analyzing robustness of networks, and searching databases for components that meet the design criteria. At the current state of synthetic biology, there are no established methods for measuring robustness or identifying components that fit a design. The same is true for databases of biological parts. TinkerCell's flexible modeling framework allows it to cope with changes in the field. Such changes may involve the way parts are characterized or the way synthetic networks are modeled and analyzed computationally. TinkerCell can readily accept third-party algorithms, allowing it to serve as a platform for testing different methods relevant to synthetic biology.
Systems level modeling of biological systems has provided mathematical explanations for experimental observations [1–3]. One of the consequences of such understanding has been the ability to design and engineer synthetic biological networks inside cells, which has given rise to the field of Synthetic Biology . The idea of engineering biological systems brings together concepts from various fields. While the laboratory procedure at present is borrowed from genetic engineering, concepts such as abstraction and interchangeable parts are taken from computer science and electrical engineering. Synthetic biology introduces the notion of biological "parts" , which are individual components that can be assembled in different ways to construct synthetic networks with different functions. Networks built by different engineers can then be reused to construct larger networks , much like a programmer using existing subroutines to build a new program more efficiently. Since synthetic biology is a young field, the best practices for making molecular biology interchangeable and programmable have not been established. Nonetheless, various success stories [6–11] have shed light on the enormous potential of synthetic biology to understand fundamental science  or create new solutions for applications ranging from fuels  to medicine .
The terminology, software, and laboratory procedures required to push synthetic biology forward are still in development. However, one can anticipate that certain key concepts such as standardization and modularity will become commonplace in synthetic biology . In synthetic biology, the need for standardization exists at different levels. At one level, there is a standard for defining parts. While biological parts are the individual components for building a network, standardized parts contain additional restrictions that are intended to make synthetic networks easier to build, more reliable, and easier to exchange. An existing standard is the standard assembly , which has made DNA assembly simpler. In future, it is anticipated that standards will also exist for describing the dynamics of a part; for example, standard promoter parts might contain a "strength" value, describing its efficiency in recruiting RNA polymerase under some standard environmental condition . This leads to the second level of standards, which describes parts in a computer-readable format such as the Resource Definition Language [17, 18], so that searching and organizing parts can be automated. Under such a framework, parts could be organized by a defined ontology. The third level of standards applies to computational models. While there are existing standard formats for representing biological models [19, 20], synthetic biology models might require additional information such as the DNA sequence or specific information about the parts that are needed to physically construct the network.
Several issues in modeling for synthetic biology have to be addressed. First, a complete model should support information for computational analysis as well building the physical biological circuit. Second, due to the fact that synthetic biology spans multiple disciplines, a large number of analyses may be possible on synthetic networks, ranging from dynamical systems analysis to analysis of the DNA sequence or statistics on the part usage in the model. The software application presented in this work, TinkerCell, will address these issues. While TinkerCell targets synthetic biology, it addresses several issues that are equally applicable outside the field. These issues include support for third-party libraries, modular design of networks, flexible modeling framework, and flexible visual format. TinkerCell is an improvement on a similar effort, Athena, by the same authors . Athena addressed issues such as modular design of networks and biological parts. However, the underlying design of Athena was not as flexible as it was intended to be. For example, Athena was not able to support an ontology of biological parts. Due to such limitations in the design, the project was completely restarted under a different name.
Existing computational tools
Numerous software tools exist that allow construction and analyses of models using scripts or visual interfaces. A comprehensive list can be found at http://www.sbml.org. Some of these applications include Jarnac , JDesigner , CellDesigner , Bio-Tapestry , PySCeS , BioJADE , little-B , SynBioSS , ProMot [29–31], and Antimony . Each of the applications have their respective advantages. For example, Jarnac and PySCeS are highly flexible due to their programming interface. Little-B is similar, building on Lisp, but it supports modules in addition. CellDesigner offers a plug-in interface, which has permitted the community to add new features. BioTapestry has a simple visual depiction for genetic networks and genetic modules. BioJADE and SynBioSS, being synthetic biology applications, support a parts database . Antimony  is a C/C++ library and an editor for defining modular models using a human-readable script. ProMot is similar but also has a visual interface that supports modularity as well as using parts to compose a circuit . We considered that TinkerCell should, at the least, contain the qualities of each, which include the flexibility of programming, plug-in support, database support, modularity, and ability to construct genetic networks.
However, the intent behind TinkerCell is not to create yet another modeling application, but to create an application that can serve as a host to the algorithms, data, and ontologies that the synthetic biology and the systems biology community can offer. Unlike most modeling applications, TinkerCell does not impose a particular modeling method (e.g. differential equations), visual representation, or a strict definition of a model. It has a very generic representation of a network, and the algorithms that use the information provide the interpretation.
Results and Discussion
The target audience for TinkerCell fall into three categories: the users, the collaborators, and the developers. The users would construct models in TinkerCell using the visual or scripting interface, access parts from database(s), and analyze the model using numerous available functions. The collaborators provide functions focusing on specific tasks. For example, one collaborator who hosts an E. coli database might add a Python script to TinkerCell for accessing that database. Another collaborator might add a complex algorithm for analyzing the stability of models. The collaborators benefit from TinkerCell because they can use it as a medium for presenting their tools to the community. The developers are experienced programmers who wish to add new interfaces or novel features to TinkerCell. Alternatively, the developers may also use the TinkerCell Core library to construct a new software application. The users benefit from both the collaborators and developers by being able to use the functions and features that they provide.
Depending on the interests of the individual, different features of TinkerCell may be more important than others. The next five sections describe four primary categories of features that are provided by TinkerCell.
Modeling and analysis
TinkerCell provides visual and script-based modeling interface, which uses Antimony scripts . The visual interface is slightly more generic than the script interface and provides default equations for transcriptional rates and other reactions. The models can also contain compartments, which are physically separated regions of a system.
Numerous functions are available in TinkerCell for analyzing a constructed model. These include deterministic  and stochastic simulation algorithms , metabolic control analysis , structural analysis , flux-balance analysis , one-dimensional and two-dimensional steady state analysis, and centrality measurements .
Third-party functions and scripts
TinkerCell provides a convenient way for users to integrate their functions with TinkerCell's user interface. The functions listed at the end of the previous section are not built-in functions. They are provided as plug-ins, or features that can be added to TinkerCell without modifying the existing program. Users with knowledge of the C or Python languages can add more functions to TinkerCell. This is done by either placing compiled C libraries in the Plug-ins/C folder or listing Python scripts in the "python-scripts.txt" text file; TinkerCell will automatically display these functions in its menu of functions.
The ability to readily add new functions will allow TinkerCell to be used as a means of contributing new functions to the community. These functions may range from mathematical analysis to database support or sequence analysis.
Models in TinkerCell are constructed using "parts" from a catalog. This catalog is defined externally in an Extensible Markup Language (XML) file; in future, this XML file will be replaced by a catalog from a parts database. Each part in the model can store a large amount of information associated with the part, such as database IDs, annotation, ontology, parameters, equations, sequence, and information required by experimentalists, such as plasmid information or restriction sites found within the part. Parts can be loaded along with all their known information from databases, although this feature depends on the growth of the databases themselves.
Being able to store information about each part permits operations such as loading of parts from a database and testing how a particular part affects the model. In contrast, a conventional modeling application that solely uses variables, equations, and parameters to describe a model will have difficulty replacing a single part, because it would need to identify the parameters and equations associated with that part. Applications that do utilize a database often limit the number of parameters in the model , thus sacrificing some flexibility. TinkerCell achieves the same without compromising flexibility.
Since the current synthetic networks are mostly composed of genetic networks, TinkerCell has extensive support for constructing gene regulatory networks by connecting parts such as promoters, ribosomal binding sites (RBS), and other genetic components listed in the parts catalog. When parts are connected together, TinkerCell will automatically derive rate equations  for transcription and translation by looking at the transcription factors bound to operator sites. The RBS part's parameters are also automatically utilized in the translation rate.
Python and C functions are available for getting and setting information about the parts. These functions can be used by third-party scripts to export information needed by experimentalists [17, 18] as well as validating the model: for example, one script can check whether the plasmid containing each part have compatible antibiotic resistance genes and restrictions sites and suggest changes. Due to limited data, such features are not present in TinkerCell, but TinkerCell's design is ideal for incorporating such features.
TinkerCell supports the ability to construct new models by connecting existing models. This is achieved by encapsulating a model as a "module". Modules contain interfaces that allow them to "connect" to other modules. This idea is borrowed from electric engineering, where complex circuits are created by connecting modules that provide specific functions. How the concept of modularity maps to biological systems is an open question [39, 40], but by supporting this idea, TinkerCell can serve as a platform for experimenting with the concept of modularity.
Plug-in interface and Core library
TinkerCell is extensible at different levels. At one level, users can add new Python and C functions to the menu of functions available in TinkerCell. At a lower level, programmers can make more significant additions by writing new plug-ins. For example, the menu of C and Python functions itself is a plug-in. Similarly, the rate equations and modeling framework is controlled by a plug-in. By adding new plug-ins, it is possible to introduce different modeling methods and new visual interfaces in TinkerCell. Users who wish to use optional plug-ins may download them into their TinkerCell Plugins folder. Detailed documentation on the plug-in framework is available at http://www.tinkercell.com, under the "Design and framework" link.
TinkerCell is organized in layers, with the bottommost layer being a generic API for constructing network drawing programs and the top-most layer being a scripting interface. TinkerCell is extensible at each of the layers. There is an additional level of flexibility provided by the XML files that load the ontology of parts and connections. The goal of this structure is to foster contributions from the community.
The Core library
Plug-ins and flexible modeling framework
TinkerCell uses the notion of plug-ins. Plug-ins are C++ programs that add new features to TinkerCell without altering the existing code, thus allowing programmers to extend TinkerCell. The current set of plug-ins are responsible for storing parameters, rate equations, and other information relevant to modeling as well as storing sequence and other details needed for physical construction of the network. The plug-ins also provide the user interface for editing these values. The complete TinkerCell application is composed of a collection of plug-ins that build on the TinkerCell Core library. Plug-ins are responsible for providing visual features such as scaling and coloring as well as modeling features such as loading the catalog of parts and connections and inserting parts and connection from that catalog.
An important role of plug-ins is defining what information can be stored in TinkerCell models. For example, the Numerical Attributes plug-in and the Stoichiometry plug-in, together, ascertain that all TinkerCell models will contain sufficient information to generate the stoichiometry matrix and rate equations, which are required for generating differential equations and stochastic simulations. Additional information, such as function definitions and events, are added by other plug-ins.
While the existing plug-ins focus on building models based on stoichiometry and rates, it is possible to write new plug-ins that focus on other modeling approaches, such as rule-based or Boolean modeling. The current set of plug-ins do not include such approaches, but the underlying structure of Tinker-Cell is not limited to any one modeling approach.
C and Python interface
All C and Python programs automatically run on separate threads, allowing a user to continue working on a model while a time consuming task is running. TinkerCell's default simulators compile the model as a separate C program, allowing the simulations to run at the speed of a compiled C program. This strategy is used by the deterministic simulator as well as the stochastic simulator, where the speed gain is more visible.
The C API has been used to include the following C programs in TinkerCell: deterministic simulation and steady state analysis using the Sundials library , stochastic simulation using a custom C library, flux balance analysis using LPsolve , and custom C programs that automatically modify reaction kinetics and calculate loops in the Jacobian matrix. The Python API has been used to include the SciPy module , PySCeS module , NerworkX module , custom programs for generating transcriptional rate formulas, and another custom program for generating FASTA files for the DNA parts in the model.
A few example Python scripts for interfacing with TinkerCell are provided (see Additional File 1).
Modular design of networks
Modules as composite parts
The synthetic biology community is familiar with the notion of a "composite part", or a biological part that has been constructed by assembling other parts together . TinkerCell can represent such an object as a module. Each sub-part inside the composite part retains its original set of attributes and other information while the composite part can have its own family and set of attributes. The C and Python API can be used to access the parameters, annotation, authorship, and list of sub-parts.
Just as with any other component in a model, modules can have rate expressions describing their input and output behavior. This allows a modeler to describe the behavior of a composite part is an abstract way without being forced to know all the internal details.
Support for standards
It is our anticipation that the growth in synthetic biology will result in a general agreement in the community about what attributes are needed to define a biological part. For example, a promoter part might be defined by its DNA sequence, operator sites, and binding affinity of RNA polymerase. The last parameter has a physical interpretation, therefore models using this parameter of the promoter part will be more coherent: the parameters used in the model will have a standard meaning that is understood by the community rather than solely by the model designer. In contrast, when a model uses parameters defined by the model designer, the physical interpretations of those parameters may not be clear. When all the parameters and rate equations in a model have clear interpretations, it becomes possible to carry out automated procedures, such as finding real parts that fit the model. The next three sections will describe how this idea of standard parameters is supported, but not enforced, in TinkerCell.
Similar to the part families, every connection between parts in TinkerCell is also identified with a family. Examples of connection families are "Biochemical reaction", "Binding reaction", and "Transcriptional regulation". Each connection family also contains attributes; for example, "Transcription regulation" contains parameters named "h", the Hill co-efficient, and "Kd", the dissociation constant. These attributes define the standard parameters required to characterize the dynamics of the reaction. Models that use such standard parameters would be easier to interpret and carry out automatic operations because the parameters have a well understood interpretations that are independent of the modeler. The connection families are also loaded from an external XML file and will be obtained directly from a database in future.
Using ontologies in models
It should be noted that the families and their attributes are not defined within TinkerCell, so the algorithms are general for any future ontology that the synthetic biology community will adopt. In addition to the family information, each part and connection in TinkerCell contains its own annotation, storing information such as authorship, references, and date. As with any other information in Tinker-Cell, all of this information is accessible and editable from C and Python.
Standard visual formats, such as the standard symbols that are used to draw electronic circuits, allow network diagrams to be unambiguous. While systems biology has made progress to standardize visual representations of biological systems , such standards for synthetic biology are still in development . TinkerCell has been designed so that it will be able to cope with a variety of visual standards or even multiple visual standards. This is achieved by having a flexible visual representation. The visual representation of proteins, genes, promoters, and the rest of the parts are stored as XML files. These files are generated using a polygon drawing program that comes with TinkerCell. The same file format is used for arrowheads and decorators such as phosphorylation sites, which keeps visual representations for those objects flexible as well.
In order to cope with a future visual standards, new files will be generated using the polygon drawing program that match the prescribed visual standard.
Since the majority of the current synthetic networks are genetic networks, TinkerCell gives special consideration to them. TinkerCell provides three different ways for modeling genetic networks, allowing the modeler to choose which is most suitable. This builtin support for multiple modeling frameworks reflects TinkerCell's underlying modeling framework, which is independent of the modeling technique.
Fractional saturation models
Using parts to model genetic networks
Explicitly defining intermediate steps
A more elaborate way of describing gene regulatory networks is to define each reaction in the transcription and translation process, including the movement of RNA polymerase across the gene . Creating such a model would be nearly impossible without some sort of automation, and TinkerCell provides a function for automating this process. While such models can avoid the use of equilibrium assumptions, simulation can be time-consuming due to the fact that each transcription and translation step is comprised of multiple reactions. But this method can often capture differences in behavior due to delays or stochastic fluctuations in the intermediate stages of transcription and translation .
List of functions provided through the C and Python interface
While the visual side of TinkerCell provides the interface for constructing a model, the analyses are carried out by C and Python functions. All of the C programs use the TC_api.h header file in order to interact with TinkerCell's visual interface. New C programs can be added to TinkerCell by building them as dynamic libraries. Python scripts can be added to TinkerCell by listing the script name and description in the "pythonscripts.txt" file. All C and Python functions automatically run on separate threads, allowing users to continue working as the program is running and utilizing the multi-core architecture of most modern computers.
The default simulators use existing C libraries that perform deterministic and stochastic simulation. For example, the stochastic simulator uses a custom Gillespie  C library available from evolvenetworks.sourceforge.net. The C library accepts a user-defined stoichiometry matrix and propensity function. The Gillespie simulator in TinkerCell takes the stoichiometry matrix and reaction rates from the TinkerCell model and generates the C code containing the stoichiometry matrix and propensity function. This C code is compiled and linked against the existing Gillespie algorithm library. The compiled C program performs the simulation and outputs the result to TinkerCell's plot window. Therefore, the simulation itself is performed by a compiled C program, which means that the simulation will be fast. Many conventional simulators interpret the rate equation string and construct an internal representation of the equation, which is used to compute the rate value during every iteration; this can be an order of magnitude slower than the compiled program. The same idea is used by the deterministic simulation, which uses the Sundials CVODE numerical integrator  to perform the simulation. One drawback is that there is a short time delay when compiling the code, which may be more than the time to simulate simple networks. Therefore, the speed gain due to this strategy is only visible for larger networks and time consuming stochastic stimulations. However, this demonstrates how existing C libraries can be used without modification to create new simulators for TinkerCell. The Tiny C Compiler  and GNU C Compiler  are used to perform the compilation in MS Windows and Unix based systems (including Mac), respectively.
Further, the simulation programs use a function called "createInputWindow" that is provided by the API. This function allows C and Python programs to create simple dialogs where the user can input parameters such as the time for simulation or the type of simulation. There are several such functions that allow C and Python programs to interact with the user through visual interfaces.
Two-parameter steady state analysis
Among the functions included with TinkerCell are one-parameter and two-parameter steady state analyses. Both functions are loaded as C plug-ins, and they use the Sundials CVODE integrator . The two-parameter steady state analysis plots a 3D surface plot, showing the value of a target variable as a function of two parameters in the system.
Visual inputs and outputs for C and Python programs
While C and Python programs are generally expected to take command-line inputs and produce command-line outputs, TinkerCell provides functions that allow C and Python functions to interact with the user directly. TinkerCell's C and Python API contain functions for displaying dialogs with input tables. The outputs from these dialogs are returned to the calling C or Python program. The C or Python program can then use the information from the model and the inputs from the user to perform some analyses. As an output, the C and Python programs can change the size or color of items in the model, circle items in the model, or display numbers or strings next to items in the model. At the same time, the programs may also print to the TinkerCell command-line window.
Flux balance analysis
The flux balance analysis is one example where a C program produces visual outputs. This particular program uses a C++ plug-in to create a custom input window. The C++ plug-in sets up the windows, tables, and buttons specifically for flux balance analysis. This window simply serves as a user interface to a C program. It receives input from the C++ plug-in and uses the LPSolve C library  to do the optimization. The C program then changes the width of the reaction arcs according to the result from LPSolve. The optimal fluxes are also displayed next to the reactions arcs (see Figure 3).
The sensitivity analysis is yet another example of visual output (see Figure 3). The senstivity converts the TinkerCell model into a PySCeS  model and uses PySCeS to perform the analysis. The script then takes the PySCeS output and colors the reactions in the network according to their control coefficients.
Accessing E. coli genetic parts through RegulonDB
RegulonDB  houses a large set of promoters, transcription factor binding sites, and RBS for the E. coli K12 strain. Through a Python script, a user can load sequences and other information such as promoter type into TinkerCell models. The script provides the user with a visual interface for selecting the parts. The script also limits the search by using known information. For example, if the transcription factor LacI is repressing a promoter, then the script will show only those binding sites that are targeted by LacI (see Figure 4), allowing a user to automatically generate the correct network using real parts. The Python script is able to use the family information of objects in the model to identify whether it is a promoter, RBS, coding region, or transcription factor. It would be impossible to search for an appropriate part with this information in the model, which demonstrates how the meta-data is useful for database search.
One of TinkerCell's objectives is to combine the appeal of visual network design with the flexibility of programming. Automatic generation of kinetics is a good example where this objective is met. There are three functions currently included with Tinker-Cell that fall in this category. First, the "Hill equation" Python function automatically generates the fractional saturation model for transcription regulation based on the activators and repressors of a promoter. Second, the "binding reactions" C program automatically generates all the intermediate stages of a protein that form multiple complexes. Third, the "multiple step process" function can automatically insert intermediate steps into any given reaction. For example, if the conversion of one molecule to another requires several intermediate stages, this function can be used to automatically generate those intermediates reactions. Gene regulatory networks are one of the areas where this function can have additional relevance.
Automatically generate multiple conformations of a protein
Consider a simple situation of two small proteins binding to a larger protein. In addition to the three proteins, there are three other complexes. However, the number of possible complexes grows exponentially with the number of binding proteins, thus modeling every complex explicitly becomes impractical . Generally, a modeler may choose to ignore such details due to the difficulty in modeling. TinkerCell offers a feature for automatically generating all necessary combinations. This feature is provided by an existing C function that recursively generates all the intermediate states and reactions. Integrating the existing function into TinkerCell required three steps: obtaining "Binding reactions" in the set of items selected by the user, getting the stoichiometry matrix for those reactions, and passing the stoichiometry matrix to the existing C function. The output from the C function is then used to modify the TinkerCell model.
Automatically generate intermediate stages of a multi-step process
Another automated feature is converting a single reaction into a multi-step reaction. This feature was inspired by the fact that transcription can be understood as several individual reactions, where each reaction is catalyzed by RNA polymerase. Modeling genetic networks in this manner can have significantly different outcomes in some situations . It would be impossible for a user to draw hundred of reactions for each transcriptional process in a model. TinkerCell offers an automated function for converting a single process into a multistep process. The visual model remains unchanged, but the selected reactions will represent multiple reactions. While an alternative approach might be delayed differential equations, this approach allows other details such as leaks in transcription and different delays based of the type of process and stochastic simulations.
TinkerCell file format
TinkerCell stores the entire model as a single XML file, which contains both the model information as well as the graphical information. All the graphical objects in TinkerCell are generated from XML files, and therefore they can be saved as XML files. The model itself is stored as a set of components and information associated with those components. The information associated with each component is a set of tables containing the parameters, reaction rates, function definitions, events, DNA sequence, annotation, and any other information pertaining to that component in the model. The file format is therefore just a list of objects, their family information, and data tables associated with the objects. The datatables are labeled as "parameters", "reaction rates", "text attributes", and so on. The format is generic, which reflects TinkerCell's underlying structure.
Example: dual feedback synthetic oscillator
Genetic circuits can be built in TinkerCell by snapping genetic parts together. The regulatory rates will be automatically assigned when promoters and RBS are connected upstream of coding segments. Delays in transcription and translation can be explicitly modeled by introducing intermediate steps automatically. These features are used effectively in the synthetic genetic circuit shown in Figure 10, which is a dual feedback oscillator that has been constructed in E. coli . Video demonstrations for the construction of this synthetic circuit and two other circuits are also provided under the "Demos" link at http://www.tinkercell.com. The videos are also available through YouTube using "TinkerCell" as the search keyword. The demonstration also shows how to load sequences from RegulonDB.
We have described an application called TinkerCell that combines the flexibility of programming with a visual interface. In addition, TinkerCell provides the structure for supporting standards and exchange of information. TinkerCell is not intended to be yet another modeling application. TinkerCell is an application for bringing together models, information, and algorithms. Therefore, it serves a host to C and Python programs that provide functions useful for biological engineering. The API has been demonstrated to be very versatile because numerous C and Python programs and packages have been readily integrated into TinkerCell.
The modular modeling framework of TinkerCell can serve two purposes. The first is to aid in the engineering process by allowing one modeler to use existing modules to design new networks. The second purpose is that modeling with modules can be used to explore the possible implication of functional modules in biology. It is an open question whether a module's functionality is retained when it is placed in different situations . Answers to such questions will be highly relevant to synthetic biology, since the ability to construct one system using another is a central theme in engineering. TinkerCell can serve as a platform for testing how different modules behave and what types of modules are able to retain their functional identity. The flexible modeling framework provides such experiments to be carried out at different levels of detail. For example, some issues regarding modularity may not be visible when using fractional saturation models of gene regulation, but they may be evident when the reactions are modeled explicitly. Since TinkerCell provides different ways of modeling, it is suitable for studying such questions.
Work in progress
TinkerCell is continually being updated. In particular, we are currently working in four areas, which include:
addition of more analysis functions from PySCeS  such as bifurcation analysis and parameter scans
better integration between the Antimony scripting language  and the visual representation to allow text and graphical views to be easily interchanged
the ability to build models with poorly defined parts.
The last feature needs more explanation. Parameters describing the complete dynamics of a given biological part are rarely known. To address this reality, TinkerCell would be able to define parameters with an upper and lower limit or a standard deviation value, indicating the uncertainty of the parameter value. The ability to include uncertainly measures on parameters will allow us to develop algorithms to determine the robustness of a variety of designs in the face of parameter variation. The most robust designs could then be selected for construction in vivo.
Integrating future synthetic biology standards  would lead to better database integration, which would allow TinkerCell to provide some useful features for experimentalists. These include:
ability to list parts that are compatible with each other, where compatibility includes physical assembly as well as known interferences or cross-talk between the parts
conveniently providing references and user ratings when they are available for parts
estimating standard measurements  on parts from experimental data and submitting the information back to a database
The above features are being developed along with database efforts.
The TinkerCell application can be successfully compiled and run on most major platforms (Windows, Mac, Linux). Installers are currently available for Windows systems. Installers for Mac will be available in the near future.
This work was partly funded by the National Science Foundation (ID 0527023- FIBR) and Microsoft's Computational Challenges in Synthetic Biology 2006 Award. Bergmann was supported by a grant from the NIH/NIGMS (GM081070).
- Hasty J, Isaacs F, Dolnik M, McMillen D, Collins J: Designer gene networks: Towards fundamental cellular control. Chaos: An Interdisciplinary Journal of Nonlinear Science 2001, 11: 207. 10.1063/1.1345702MATHView ArticleGoogle Scholar
- Isaacs F, Hasty J, Cantor C, Collins J: Prediction and measurement of an autoregulatory genetic module. Proceedings of the National Academy of Sciences 2003,100(13):7714-7719. 10.1073/pnas.1332628100View ArticleGoogle Scholar
- Ozbudak E, Thattai M, Lim H, Shraiman B, van Oudenaarden A: Multistability in the lactose utilization network of Escherichia coli. Nature 2004,427(6976):737-740. 10.1038/nature02298View ArticleGoogle Scholar
- Endy D: Foundations for engineering biology. Nature 2005, 438: 449-453. 10.1038/nature04342View ArticleGoogle Scholar
- Purnick P, Weiss R: The second wave of synthetic biology: from modules to systems. Nature Reviews Molecular Cell Biology 2009,10(6):410-422. 10.1038/nrm2698View ArticleGoogle Scholar
- Elowitz M, Leibler S: A synthetic oscillatory network of transcriptional regulators. Nature 2000,403(6767):335-338. 10.1038/35002125View ArticleGoogle Scholar
- Gardner T, Cantor C, Collins J: Construction of a genetic toggle switch in Escherichia coli. Nature 2000,403(6767):339-342. 10.1038/35002131View ArticleGoogle Scholar
- Entus R, Aufderheide B, Sauro H: Design and implementation of three incoherent feed-forward motif based biological concentration sensors. Systems and Synthetic Biology 2007,1(3):119-128. 10.1007/s11693-007-9008-6View ArticleGoogle Scholar
- Stricker J, Cookson S, Bennett M, Mather W, Tsimring L, Hasty J: A fast, robust and tunable synthetic gene oscillator. Nature 2008,456(7221):516-519. 10.1038/nature07389View ArticleGoogle Scholar
- Tigges M, Marquez-Lago T, Stelling J, Fussenegger M: A tunable synthetic mammalian oscillator. Nature 2009,457(7227):309-312. 10.1038/nature07616View ArticleGoogle Scholar
- Tabor J, Salis H, Simpson Z, Chevalier A, Levskaya A, Marcotte E, Voigt C, Ellington A: A Synthetic Genetic Edge Detection Program. Cell 2009,137(7):1272-1281. 10.1016/j.cell.2009.04.048View ArticleGoogle Scholar
- Lee S, Chou H, Ham T, Lee T, Keasling J: Metabolic engineering of microorganisms for biofuels production: from bugs to synthetic biology to fuels. Current Opinion in Biotechnology 2008,19(6):556-563. 10.1016/j.copbio.2008.10.014View ArticleGoogle Scholar
- Anderson J, Clarke E, Arkin A, Voigt C: Environmentally Controlled Invasion of Cancer Cells by Engineered Bacteria. Journal of Molecular Biology 2006, 355: 619-627. 10.1016/j.jmb.2005.10.076View ArticleGoogle Scholar
- Heinemann M, Panke S: Synthetic biology-putting engineering into biology. Bioinformatics 2006,22(22):2790-2799. 10.1093/bioinformatics/btl469View ArticleGoogle Scholar
- Shetty RP, Endy D, Knight TFJ: Engineering Bio-Brick vectors from BioBrick parts. J Biol Eng 2008, 2: 5. 10.1186/1754-1611-2-5View ArticleGoogle Scholar
- Kelly JR, Rubin AJ, Davis JH, Ajo-Franklin CM, Cumbers J, Czar MJ, de Mora K, Glieberman AL, Monie DD, Endy D: Measuring the activity of BioBrick promoters using an in vivo reference standard. J Biol Eng 2009, 3: 4. 10.1186/1754-1611-3-4View ArticleGoogle Scholar
- Galdzicki M, Chandran D, Nielsen A, Morrison J, Cowell M, Grünberg R, Sleight S, Sauro H: BBF RFC 31: Provisional BioBrick Language (PoBoL). 2009.Google Scholar
- Synthetic Biology Open Language[http://openwetware.org/wiki/The_BioBricks_Foundation:Standards/Technical/Exchange]
- Hucka M, Finney A, Sauro H, Bolouri H, Doyle J, Kitano H, Arkin A, Bornstein B, Bray D, Cornish-Bowden A, et al.: The systems biology markup language (SBML): a medium for representation and exchange of biochemical network models. Bioinformatics 2003,19(4):524-531. 10.1093/bioinformatics/btg015View ArticleGoogle Scholar
- Lloyd C, Halstead M, Nielsen P: CellML: its future present and past. Progress in Biophysics And Molecular Biology 2004, 85: 433-450. 10.1016/j.pbiomolbio.2004.01.004View ArticleGoogle Scholar
- Chandran D, Bergmann F, Sauro H: Athena: Modular CAM/CAD Software for Synthetic Biology. eprint arXiv: 0902.2598 2009.Google Scholar
- Bergmann F, Sauro H: SBW-a modular framework for systems biology. Proceedings of the 38th conference on Winter simulation, Winter Simulation Conference 2006, 1637-1645.Google Scholar
- Funahashi A, Morohashi M, Kitano H, Tanimura N: CellDesigner: a process diagram editor for gene-regulatory and biochemical networks. Biosilico 2003, 1: 159-162. 10.1016/S1478-5382(03)02370-9View ArticleGoogle Scholar
- Levine M, Davidson E: Gene regulatory networks for development. Proceedings of the National Academy of Sciences 2005,102(14):4936-4942. 10.1073/pnas.0408031102View ArticleGoogle Scholar
- Olivier B, Rohwer J, Hofmeyr J: Modelling cellular systems with PySCeS. Bioinformatics 2005,21(4):560-561. 10.1093/bioinformatics/bti046View ArticleGoogle Scholar
- Goler J: BioJADE: A Design and Simulation Tool for Synthetic Biological Systems. Technical report 2004.Google Scholar
- Mallavarapu A, Thomson M, Ullian B, Gunawardena J: Programming with models: modularity and abstraction provide powerful capabilities for systems biology. Journal of The Royal Society Interface 2009,6(32):257-270. 10.1098/rsif.2008.0205View ArticleGoogle Scholar
- Hill A, Tomshine J, Weeding E, Sotiropoulos V, Kaznessis Y: SynBioSS: the synthetic biology modeling suite. Bioinformatics 2008,24(21):2551. 10.1093/bioinformatics/btn468View ArticleGoogle Scholar
- Ginkel M, Kremling A, Nutsch T, Rehner R, Gilles E: Modular modeling of cellular systems with ProMoT/Diva. Bioinformatics 2003,19(9):1169-1176. 10.1093/bioinformatics/btg128View ArticleGoogle Scholar
- Marchisio M, Stelling J: Computational design of synthetic gene circuits with composable parts. Bioinformatics 2008,24(17):1903. 10.1093/bioinformatics/btn330View ArticleGoogle Scholar
- Mirschel S, Steinmetz K, Rempel M, Ginkel M, Gilles E: PROMOT: modular modeling for systems biology. Bioinformatics 2009,25(5):687. 10.1093/bioinformatics/btp029View ArticleGoogle Scholar
- Smith L, Bergmann F, Chandran D, Sauro H: Antimony: A modular model definition language. Bioinformatics 2009,25(18):2452-2454. 10.1093/bioinformatics/btp401View ArticleGoogle Scholar
- Registry of Standard Biological Parts[http://partsregistry.org/]
- Hindmarsh A, Brown P, Grant K, Lee S, Serban R, Shumaker D, Woodward C: SUNDIALS: Suite of nonlinear and differential/algebraic equation solvers. ACM Transactions on Mathematical Software (TOMS) 2005,31(3):363-396. 10.1145/1089014.1089020MATHMathSciNetView ArticleGoogle Scholar
- Gillespie D, et al.: Exact stochastic simulation of coupled chemical reactions. The Journal of Physical Chemistry 1977, 81: 2340-2361. 10.1021/j100540a008View ArticleGoogle Scholar
- Berkelaar M, Eikland K, Notebaert P, et al.: lpSolve: Open Source (Mixed-Integer) Linear Programming System. Eindhoven U. of TechnologyGoogle Scholar
- Hagberg A, Schult D, Swart P: NetworkX Library developed at the Los Alamos National Laboratory Labs Library (DOE) by the University of California.2004. [https://networkx.lanl.gov]Google Scholar
- Shea M, Ackers G: The OR control system of bacteriophage lambda. A physical-chemical model for gene regulation. J Mol Biol 1985, 181: 211-30. 10.1016/0022-2836(85)90086-5View ArticleGoogle Scholar
- Del Vecchio D, Ninfa A, Sontag E: Modular cell biology: retroactivity and insulation. Molecular Systems Biology 2008, 4: 161-177. 10.1038/msb4100204View ArticleGoogle Scholar
- Sauro H: Modularity defined. Molecular Systems Biology 2008, 4: 166-168. 10.1038/msb.2008.3View ArticleGoogle Scholar
- Qt - a cross-platform application and UI framework[http://www.qtsoftware.com/products/]
- Jones E, Oliphant T, Peterson P, et al.: SciPy: Open source scientific tools for Python.2001. [http://www.scipy.org]Google Scholar
- Sauro H, Penrodyn P: A Biochemical "NAND" Gate and Assorted Circuits. In Modern Trends in Biothermokinetics: Proceedings of the Fifth International Meeting Held in Bordeaux-Bombannes, France, September 23-26. 1992. Plenum Publishing Corporation; 1993:133.View ArticleGoogle Scholar
- Gama-Castro S, Jimenez-Jacinto V, Peralta-Gil M, Santos-Zavaleta A, Penaloza-Spinola M, Contreras-Moreira B, Segura-Salazar J, Muniz-Rascado L, Martinez-Flores I, Salgado H, et al.: RegulonDB(version 6. 0): gene regulation model of Escherichia coli K-12 beyond transcription, active(experimental) annotated promoters and Textpresso navigation. Nucleic Acids Research 2008,36(1):120-124. 10.1093/nar/gkn491Google Scholar
- Le Novère N, Hucka M, Mi H, Moodie S, Schreiber F, Sorokin A, Demir E, Wegner K, Aladjem M, Wimalaratne S, et al.: The Systems Biology Graphical Notation. Nature Biotechnology 2009,27(8):735-741. 10.1038/nbt.1558View ArticleGoogle Scholar
- BioBrick Open Graphical Language[http://openwetware.org/wiki/Endy:Notebook/BioBrick_Open_Graphical_Language]
- Kosuri S, Kelly J, Endy D: TABASCO: A single molecule, base-pair resolved gene expression simulator. BMC bioinformatics 2007, 8: 480. 10.1186/1471-2105-8-480View ArticleGoogle Scholar
- Bellard F: TCC: Tiny C compiler.[http://fabrice.bellard.free.fr/tcc]
- GCC, the GNU Compiler Collection[http://gcc.gnu.org/]
- Blinov M, Faeder J, Goldstein B, Hlavacek W: BioNet-Gen: software for rule-based modeling of signal transduction based on the interactions of molecular domains. Bioinformatics 2004,20(17):3289. 10.1093/bioinformatics/bth378View ArticleGoogle Scholar
- Sauro H, Hucka M, Finney A, Wellock C, Bolouri H, Doyle J, Kitano H: Next Generation Simulation Tools: The Systems Biology Workbench and BioSPICE Integration. Omics A Journal of Integrative Biology 2003, 7: 355-372. 10.1089/153623103322637670View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.