The Ras Executable Model (REM) is a model of the signaling pathway of the Ras family proteins (K-RAS, N-RAS, H-RAS), including their regulators and effectors, at a biochemical level of detail. It is a rule-based model written in PySB.
REM consists of three interlinked levels. Model components consist of PySB modules that implement pathway mechanisms. A model scenario corresponds to a specific use case and imports one or more model components in a fit-to-purpose manner to instantiate an executable model. Each model scenario can have a set of corresponding model analysis scripts.
REM is built via a combination of automated assembly using INDRA (Integrated Network and Dynamical Reasonging Assembler), and manual development by a team of modelers.
The Ras Executable Model is being developed as part of DARPA’s Big Mechanism program by the Sorger Lab at Harvard Medical School and collaborators within the Big Mechanism program. In the context of the program’s overall effort to automate the construction of models from the scientific literature, the development of the Ras model has the following motivations:
Systems biology resource.
Opportunity to engage biological domain experts.
Investigation of the combination of manual and automated reading and modeling process.
Resource for improving reading algorithms.
Reference standard for evaluating reading algorithms.
A “literate” model¶
To better understand the relationship between scientific texts and executable models, in REM the documentation and evidence supporting the various biochemical reactions is a primary, rather than a supporting feature. As a means of more fully integrating textual evidence with executable modeling code we have adopted a “literate” style of programming. As first described by computer scientist Donald Knuth, in literate programming,
Instead of imagining that our main task is to instruct a computer what to do, let us concentrate rather on explaining to human beings what we want a computer to do.
Our goal for REM is for a human reader with biological and/or modeling expertise to read it, much like a scientific paper or review, and evaluate whether the implementation of the model’s various mechanistic assertions are adequately supported by the textual evidence.
To do this, we exploit the fact that in PySB, biological models are implemented as Python programs, which allows tools and conventions for computer programming to be applied to the organization and documentation of models. In REM, we embed the PySB modeling code within the documentation itself, which is written in reStructuredText. The documentation is processed and formatted using the Python documentation generator, Sphinx, while the remainder of the build process extracts the model code from the documentation, and executes it to generate various outputs: exports of the model in several formats (BNGL, Kappa, SBML), visualizations, and simulation results.
Ty Thomson’s Yeast Pheromone Model is an excellent prior example of this general approach.
To generate the text of the documentation, Sphinx is the only dependency, which can be installed via Pip with:
pip install -U Sphinx