Computational Morphodynamics Modeling Project
Results
A major goal of this project was to explore options for future development of the Computable Plant Group's Sigmoid Modeling Environment.
a) Remote data sourcing. Once a model has been developed, biologists may want to run it on alternate data sets. In the Mathematica implementation of the Wuschel model,cellular coordinates were specified as local data files. While this has utility, we wanted to develop a more general-purpose method for importing geometries into the modeling environment. We made use of the RESTful API provided by the Bisque (Bio-Image Semantic Query User Environment) system to access and retrieve photographs of shoot apical meristems and data points that were programmatically assigned to represent cell centers. This was done transparently to the consumer of our own RESTful PI and additional data sources could be easily added. We considered, but did not implement, passing the rendered results of simulation sessions to the Bisque database.
b) Communication via modern, standard protocols. The current API for interacting with the Sigmoid Middleware, which is a web service for interacting with and performing biomolecular simulation,is implemented in SOAP. While this is appropriate technology for the task, it is fairly opaque to third-party implementers who might hope to create a new interface to the middleware application.We developed a simple, self-documenting API for driving a 2D biomolecular modeling application. The web service provides human-readable documentation at the root level of the service. Command structures are implemented as subject-predicate pairs, with subjects returning an annotated list of potential predicates when called invoked alone.
c) Scalable code execution. Complex biomolecular simulations may be time- and resource-intensive. While Mathematica has a grid-enabled version, it is proprietary and not inexpensive. Instead, we aimed to explore distributed execution and session queing by passing Mathematica code into a cluster environment. We chose Condor for its support of heterogenous clusters and ability to dynamically stage dependencies (such as shared libraries and data files). The end result was that our prototype CGI scaled immediately to the number of cluster nodes in the UA system with available Mathematica installations.
d) AJAX-based UI. As an exercise, we used the Yahoo User Interface Library (YUI) to build a HTML/AJAX user interface from the self-documenting API to test functionality and robustness of the application and interface. An alternative interface, such as one built in Flash or a command-line executable, could easily be developed.
Deliverables
Sample Web Application
A demo version of the UI+CGI combination is available at http://sigmoid.iplantcollaborative.org/
If the simulation does not work, contact us and while you await a reply, feel free to check out the already-completed sessions.
The RESTful API upon which this service is built is documented at http://sigmoid.iplantcollaborative.org/simul8/bin/mathsrvr.cgi/=
We've also provided some demo movies of the application.
- A biochemical pathway, intercellular network, and cellular geometry are selected as the basis for a modeling session [Movie 1]
- Results of a simulation session are examined [Movie 2]
- The BISQUE system is used to determine nuclear centers from a micrograph [Movie 3]
- The results from the BISQUE-catalyzed analysis are used in a simulation session [Movie 4]
- The results from the BISQUE-sourced simulation are examined [Movie 5]
Source Code
We are pleased to offer a read-only Subversion repository containing source code for the HTML/AJAX UI, the mathsrvr experimental modeling CGI, and a first-pass unified source tree for the Sigmoid Modeling Environment. Using the command-line Subversion client, you may check out the repository as follows:
svn checkout "http://svn.iplantcollaborative.org/pre_projects/compmorph/trunk"
If you're using Windows, we suggest you use the TortoiseSVN client to access this URI.
Contributors
Ben Compani, UCI; Boguslaw Obara, UCSB; Bruce Shapiro, Caltech; Edwin Skidmore, UA; Eric Mjolsness, UCI; Jianlin Cheng, UMo; Liya Wang, CSHL; Matthew Vaughn, CSHL; Sheldon McKay, CSHL; Zheng Wang, UMo