 In-vivo and in-silico dictyostelium
development. Single-celled organisms self-organize into a
multicellular blob. The blob forms a stalk with spores at the
end that disperse for reproduction. The inset shows a Cellular
Potts Model-based simulation of the spore formation. (images
from dictyBase)
|
OSL/Biocomplexity Institute Collaboration
Since 2003, the OSL has been collaborating with the Biocomplexity
Institute at Indiana University.
The main research focus at the Biocomplexity Institute is the
creation of computational models of chicken-limb development
using the
Cellular Potts Model (CPM), an extension to the
Potts
model for (biological) cell-level modeling. Using data
gathered from in-vivo experiments, scientists have
successfully modeled complex behaviors such as cell-sorting and
chemotaxis.
My main focus in this collaboration has been
researching and developing software and support infrastructures
that allow the scientists to focus on science, not software
engineering.
Over the course of the last two years, we helped expand the scope of the resources available to the local scientists. One key aspect of this was identifying university resources that can contribute directly to the development process and helping to manage these relationships. This has led to ongoing partnerships with both the AVL and SDAL at IUPUI for visualization and high-performance computing support.
In the summer of 2004, we began to move away from the home-grown code traditionally developed by scientists to CompuCell3D, a CPM infrastructure developed by the LCLS at Notre Dame. The LCLS and Biocomplexity Institute are both partially funded under the same grant and CompuCell3D is a successful result of that funding. In order to support the transition, I extended CompuCell3D to support a Python scripting interface for rapid prototyping and visualization. I also worked closely with the LCLS by providing architectural feedback and future requirements for CompuCell3D. Recently, we hired a full-time developer to take over most of the development tasks at IU.
An overview of our research collaboration is included in this poster:
Links to CompuCell3D educational and development material:
Links to visualization examples:
Artifacts from the early days of the collaboration:
Over the course of the last two years, we helped expand the scope of the resources available to the local scientists. One key aspect of this was identifying university resources that can contribute directly to the development process and helping to manage these relationships. This has led to ongoing partnerships with both the AVL and SDAL at IUPUI for visualization and high-performance computing support.
In the summer of 2004, we began to move away from the home-grown code traditionally developed by scientists to CompuCell3D, a CPM infrastructure developed by the LCLS at Notre Dame. The LCLS and Biocomplexity Institute are both partially funded under the same grant and CompuCell3D is a successful result of that funding. In order to support the transition, I extended CompuCell3D to support a Python scripting interface for rapid prototyping and visualization. I also worked closely with the LCLS by providing architectural feedback and future requirements for CompuCell3D. Recently, we hired a full-time developer to take over most of the development tasks at IU.
An overview of our research collaboration is included in this poster:
- Research Collaboration Poster
- Support Movies (tar.gz)
Links to CompuCell3D educational and development material:
- CompuCell Introduction
- UML Inheritance Diagram
- Config file example (xml)
- CompuCell3D Python Prototype
Links to visualization examples:
- Charlie Moad's vis prototypes (go to projects->Potts Cellular Simulation) (html)
- Vis Prototypes (html)
- Chemical Concentration (html)
Artifacts from the early days of the collaboration:
- Limb 3D pseudo-UML (html)
- Potts 2.0 Requirements Notes
- Limb 3D Source Code and Data (html)
- Biocomplexity SourceGrid (sg.osl.iu.edu) (login required)
- CompuCell3D Python Prototypes (tar.gz)
Typed Diffusion Limited Aggregation (Typed DLA)
Diffusion limited aggregation (DLA) is a technique that models
random particle aggregation. In the basic algorithm, a single
particle is released on a random walk and stops when it
collides with an existing stationary particle. The resulting
structures are similar to many patterns found in nature. In
this project, we extend the DLA model to include 'typed'
particles. If two incompatible particles collide, the moving
particle continues its walk. We measure the effect of typed
particles by measuring the fractal dimension of the resulting
images. This work was done as the course project for P700 -
Fractals and Pattern Formation.
- Typed DLA Presentation
- Typed DLA Paper