Cell sorting

Biocellion Studies

Timings These are results from Testing Biocellion on the machine with the following details: CPU: Intel(R) Core(TM) i7-3770K CPU @ 3.50GHz Using: 4 threads Ram: 16GB

This was a test of the Cell Sorting Model speed with an increasing number of agents. This was a test of the Yeast Model speed with an increasing number of frames.(agent count increased with time) Cell Sorting

A test for the cell sorting model was conducted using a low, medium, and high levels of adhesion.

The model ran for 20 frames using all 27 combinations of the following adhesions: high: 3 medium: 1.5 low: .001

The original sorting model we had used AA adhesion of 2.5, BB adhesion of 2.5, and AB BA adhesion as .001

Morphology results for alternative adhesion levels can be found here.

Cell Count and Growth Using the newest version of Biocellion available to us,( 12/31/12) we ran the default parameter yeast model and had the following results for cell growth.

All data was collected manually from Paraview.

We will be refining the model based on the cell growth curves. Is there a setting within the model where it would be possible output this data in a simple text file? We need this to automate the process of scoring the model against the data.

I think Paraview has a feature for this. You can run some statistics filters using Paraview.

Growth Animations We have recorded a few animations of our Yeast Run that has the information recorded above.

Here is an angled animated view of the yeast growth http://youtu.be/H0rdtXswetg

Here is a Cross Section animated view of the Yeast growth http://youtu.be/ChFf9OVBSvM We are setting up to shape features of the morphology to further improve the model and validate the model. Unfortunately the Paraview files don’t contain all of the information we need in a readable format. Is there a setting that will produce simple text file data details the position and state of every particle in the system? Without this we will be unable to automate the matching of simulated and real morphologies since we need to extract shape features such as height, width, agar penetration etc.

Paraview has all the different types of filters and data processing capabilities. I think those are something you may want to see first.

Solute Visualization Glucose Concentration cross section: Animation: http://youtu.be/nbldGYCol7c Screenshot at timestep 3000

Ethanol Concentration: Animation: http://youtu.be/tC_BNuhwfek Screenshot at timestep 5000

Video cross section of growth and glucose concentration: http://youtu.be/QRlV8k1sAdY

The glucose and ethanol concentration files appear to have the axis rotated? I’m unsure of where this can be adjusted within the model files or whether it needs to be changed within the biocellion build.

It appears that the grid output in the concentration visualization files has a much lower resolution than the actual cells of the model.

Is this the same grid that is used in the simulation? This may be the cause of the apparent grid pattern in cell distribution. In order to refine the model, is it possible to change these values within the xml, header, or .cpp files?

The grid data should have the same axis setup with the agent data, and I will look for those. The grid spacing is determined by d_max in the simulation setup (d_max should be larger than the maximum mechanical interaction distance between any two particles).