Biocharts: a visual formalism for complex biological systems

Hillel Kugler, Antti Larjo, David HarelJournal of The Royal Society Interface 2009 · DOI

TL;DR

This paper introduces Biocharts, a novel two-tier visual modeling language designed to bridge the gap between high-level biological system behavior and low-level molecular dynamics. By combining statecharts for system-level behavior with other formalisms for detailed pathway simulation, Biocharts aims to create fully executable models that are intuitive for biologists to use and understand.

What Problem Does This Paper Address?

Understanding complex biological systems, like cellular networks and pathways, is a monumental task. A key challenge is integrating information across different scales – from intercellular communication down to the intricate dance of molecules within a cell. Current visual modeling approaches often struggle with the sheer volume of data and the complexity of representing dynamic interactions. They can be limiting when dealing with numerous excitatory or inhibitory arrows and lack a natural way to express multi-level system behavior. This makes it difficult for biologists to actively participate in computational modeling and for researchers to build comprehensive, executable models that can be used for in silico experiments and predictions.

Methods

The core of the proposed solution is Biocharts, a two-tier compound visual language.

  • High-Level Behavior: This tier utilizes statecharts, a well-established visual formalism from software and systems engineering known for its success in modeling complex reactive systems. Statecharts allow for the representation of system states, transitions between states triggered by events, and hierarchical organization of behavior.
  • Low-Level Dynamics: This tier is where the detailed molecular pathways and network dynamics are specified. Biocharts allows for the integration of any appropriately well-defined language for this purpose, preferably a diagrammatic one. The authors suggest using existing tools and formalisms for this, such as:
  • STOCHSIM: For stochastic simulations of molecular pathways (e.g., bacterial chemotaxis signaling).
  • Flux Balance Analysis (FBA): For modeling metabolic networks, assuming a steady state.
  • Microsoft Solver Foundation (MSF08): To solve the linear programming problems associated with FBA.

The key innovation is how these two tiers are connected: the statecharts act as controllers, invoking the lower-level models based on the current state of the system. This creates a fully executable model where the high-level behavior dictates when and how the detailed simulations are run.

Datasets

Not discussed in the paper. The authors use a well-studied biological system (bacterial chemotaxis) and existing models for its components, rather than relying on specific datasets for training or evaluation.

Key Insights and Ideas

  • Bridging the Scale Gap: The most significant insight is the elegant solution for integrating high-level system behavior with low-level molecular detail. By using statecharts as the orchestrator, Biocharts provides a natural way to manage complexity and ensure that different levels of biological organization interact coherently.
  • Biologist-Centric Design: A primary goal is to empower biologists. The visual nature of statecharts and the ability to integrate familiar diagrammatic notations for pathways aim to make computational modeling more accessible and intuitive for domain experts.
  • Executable Models for In Silico Experiments: The emphasis on creating fully executable models is crucial. This allows for rigorous simulation, analysis, and prediction, moving beyond static diagrams to dynamic, interactive representations of biological systems.
  • Modularity and Abstraction: The two-tier approach inherently promotes modularity. Different biological processes can be modeled as separate modules (e.g., chemotaxis, metabolism) and integrated within the statechart framework, allowing for abstraction and easier management of complexity.

Key Results

The paper demonstrates the Biocharts approach using a model of bacterial chemotaxis in Escherichia coli, integrating both the signaling pathway for chemotaxis and the cell's metabolic processes.

  • Integrated Model: They successfully built a model that combines the high-level behavior of bacterial movement (run/tumble states) with the low-level molecular dynamics of chemotaxis signaling (using STOCHSIM) and metabolic activity (using FBA and dFBA).
  • Executable Simulation: The combined model is fully executable, allowing for simulations of bacterial population dynamics, including phenomena like band formation driven by chemotaxis and metabolism.
  • Visual Representation: Figures illustrate the statechart structure of the bacterium, its internal components, and the graphical output of simulated bacterial movement and environmental aspartate concentration.
  • Demonstration of Capabilities: The model showcases the ability to integrate different aspects of bacterial behavior (chemotaxis, metabolism, cell division) into a coherent, quantitative system-level model.

Limitations and Future Work

  • Scope of Current Model: The authors note that their current modeling effort focuses on single bacteria or small populations (up to 100 bacteria).
  • Computational Complexity: Simulating larger populations or more complex systems with Biocharts can be computationally intensive, especially with the added pathway-related computations.
  • Future Work:
    • Developing a generic tool to support the Biocharts approach.
  • Exploring more abstract representations of behavior for efficiency.
  • Caching computation results and distributing models across clusters for scalability.
  • Investigating the use of Biocharts with scenario-based languages (like live sequence charts) for future research directions in modeling languages.
  • Demonstrating that these emerging global models are quantitatively accurate and can make new, experimentally verifiable predictions.

Summary updated

Abstract

We address one of the central issues in devising languages, methods and tools for the modelling and analysis of complex biological systems, that of linking high-level (e.g. intercellular) information with lower-level (e.g. intracellular) information. Adequate ways of dealing with this issue are crucial for understanding biological networks and pathways, which typically contain huge amounts of data that continue to grow as our knowledge and understanding of a system increases. Trying to comprehend such data using the standard methods currently in use is often virtually impossible. We propose a two-tier compound visual language, which we call Biocharts, that is geared towards building fully executable models of biological systems. One of the main goals of our approach is to enable biologists to actively participate in the computational modelling effort, in a natural way. The high-level part of our language is a version of statecharts, which have been shown to be extremely successful in software and systems engineering. The statecharts can be combined with any appropriately well-defined language (preferably a diagrammatic one) for specifying the low-level dynamics of the pathways and networks. We illustrate the language and our general modelling approach using the well-studied process of bacterial chemotaxis.
BibTeX
@article{kugler2009biocharts,
  author = {Hillel Kugler and Antti Larjo and David Harel},
  title = {{Biocharts: a visual formalism for complex biological systems}},
  journal = {Journal of The Royal Society Interface},
  year = {2009},
  doi = {10.1098/rsif.2009.0457},
}
Figure