Multi-scale Bio-Modeling and Visualization 

Project Notes: Modeling the Neuro-Muscular Junction

Last Update


In this project we want to model the neuro-muscular junction (see Figure  1). By modeling we mean properly placing the entities involved in the physiological activity. In essence, it is a geometric modeling exercise and for the time being we will focus on static modeling, i.e., we will consider a snapshot and place the molecules - Acetylcholine receptor (AchR) on the thickened cell membrane and Acetylcholinesterase (AchE) in the synaptic cleft which is the volume entrapped between the nerve cells and the cell membranes (see Figure  2 for an example result). For a spatially realistic geometric modeling of this biological system we rely on the following resources.



Figure 1. Left: A sketch of neuro-muscular junction. Right: Mesh model of the cell membrane.



Figure 2. An example showing the placement of the molecules (AchR and AchE) in NMJ (from

1. Geometric Model

·                    Mesh model of neuro-muscular junction:  We shall use the mesh model of a part of the neuro-muscular junction provided by Dr. Tom Bartol of SALK Inst. The visualization of the mesh can be seen in Figure  1. The mesh can be downloaded from The lower contorted surface is of cell membrane and the volume enclosed by the two is the synaptic cleft.

·                    Molecules:  There are two sets of molecules we need to deal with in this modeling exercise. First set consists of the AchR molecules (PDBID 2BG9) and the second set consists of AchE (PDBID 1C2B).

The goal is to

1.                  Devise an automatic way of placing the AchR molecules in the thickened surface of the cell membrane. The parameters are the densities of these molecules and their orientations which might differ from molecule to molecule.

2.                  Devise an automatic way of placing the AchE molecules in the synaptic cleft. Again, the parameters guide the density and orientations of these molecules.

In the next section we list some of the detailed steps and the routines/algorithms that one might want to use/design to create the spatially realistic model of a neuro-muscular junction.

2. Algorithms and Software

One possible way to populate the molecules in their respective regions is to use the following (high-level) sketch. We identify different modules that one needs to develop to accomplish the goal. There could be variants of each step and ofcourse one might possibly think of a completely different way to do the task. Therefore the following guideline is rather indicative of a possible algorithm with necessary components than a lab instruction manual.

1. Signed Distance Function representation of Geometry

Given the mesh of neuro-muscular junction, create a grid volume enclosing the mesh where every grid vertex has a scalar value that denotes its distance from the mesh. The function sampled at these vertices is the distance function. One can add a sign with the function value where  indicates that the point is outside the volume enclosed by the mesh and  indicates that it is inside.

While this produces a volume representation of the mesh, from which the mesh can be re-extracted as the -set by any available contouring method, this doesn’t quite serve the purpose for the geometry of the molecules obtained from Protein Data Bank. There the geometry is a schematic representation of the atoms and bonds in between. To convert them into signed distance volume one needs to do an extra step of blurring the molecule using Gaussian kernel and take the level set which produces a smooth approximation of the molecule. From the meshed level set again one can use the SDF library to generate a volume of signed distance function.

1. Routines

·                    Blur:  Use Exercise 1.

·                    SDF:  The library is called SDF Lib and it can be downloaded from

2. Thickening of Cell Membrane

Once the SDF volume of the cell membrane is created it is possible to obtain a thickness of the cell membrane geometry by taking an interval volume in the grid where one isovalue corresponds to the membrane geometry and the other isovalue produces a level set offset by a parameter from the membrane surface. The interval volume therefore corresponds to a thickness around the membrane surface. The second parameter of the interval volume guides the amount of thickness.

1. Routines

LBIE-Mesher produces an interval volume from a scalar volume.

3. Molecule Placement

In the above two steps we have created a suitable computer representation of the platform where the placement procedure can start. This process has two distinct parts - one for the placement of AchR in the thickened membrane and the other for the placement of AchE in the synaptic cleft. However the algorithm used is same and therefore we list that only under a single subsection.

Placement of a single molecule is guided by a set of parameters - Rotational and Positional. Following this set of parameters one can orient the small SDF cube (volume) of a molecule. Now the task is to penetrate the target volume and actually place it which requires a modification in the SDF volume of the neuro-muscular junction to correctly reflect the presence of the molecule so that if one performs a contouring of the level set one gets the isosurface which indicates the protruding molecule at the desired position and with the desired orientation.


Figure 3. From left to right: AchR molecule, its placement parameters (intra-extra cellular extents in the cell membrane) and one example embedding.

Once the placement of one molecule is achieved, the process can be repeated to place the whole set where the density is guided by another parameter. The paremeters required to place the molecules (AchR, AchE) are described below. These parameters are crucial to obtain a spatially realistic model of the whole system. So do read them carefully. More details can be found in [4](Section 4.5, page 117).

·                    Orientation of AchR:  The AchR molecules should be oriented in such a way that they are normal to the surface of the cell membrane. As can be seen from Figure  3 these molecules are elongated in one direction and that direction should be perpendicular to the membrane surface. The intra/extra cellular extent is also shown in Figure  3.

·                    Density of AchR:  There are total 91,000 AchR to be placed on the membrane surface. The density of AchR is not same though on different parts of the membrane geometry. We divide the geometry into two parts - top and middle. On the top part there will be total of 66,000 molecules and in the gorge (middle)there are total 25,000 molecules.



Figure 4. Basal Lamina: A permeable membrane near the cell membrane where the tetrameric units of AchE are to be placed. Right top figure shows a schematic diagram of one such tetrameric unit. It has a tail which punctures the cell membrane and goes inside. Right bottom figure shows a isosurface representation of the same color coded according to electrostatic properties.

1.                  Density of AchE:  The density of AchE is however same all throughtout the basal lamina. There are total 59,000 AchE molecules distributed in this region. The basal lamina is situated near the cell membrane surface. For visualization, see Figure  4.

1. Routines

You need to devise an algorithm to do this part. We list some high-level implementation ideas that one might follow.

Every voxel in the SDF volume of the neuro-muscular junction can be flagged following their containment in the boundary of the cell membrane, the thick interval volume around the cell membrane surface, the basal lamina and the synaptic cleft. Create a datastructure that holds the small SDF volume of a molecule along with its other attributes, e.g. positional attributes etc. Each voxel then can have a list of subvolumes for every molecule that is placed within the confinement of that voxel. As we mentioned earlier, the function values at the vertices need to be modified after each placement and this can be done locally.

4. Visualization: Level of Detail Exploration

Although listed at the end of this sketch, this part should be used to verify the correctness of your implementation in each step. We encourage that you use TexMol for the visualization. It has a very useful feature that will let you perform Level-of-Detail (LOD) visualization of the model that you build. This part will have more details after Vinay and Powei finishes incorporating their LOD viewer into TexMol.

5. Useful Tools and Libraries

You are encouraged to use some or all of the tools and libraries listed here for this project. These tools are all available from CCV Useful Links page. This list will be updated as we identify some more tools and routines available that can be useful in this process. Also we will give you more references to learn the functionalities and working principle of these tools in detail.

·                    SDF Lib

·                    Blurrng routine [1]

·                    VolRover

·                    TexMol [1]

·                    LBIE-Mesher [5,6]

·                    Contour Spectrum [3]

·                    Fast Isocontour [2]

3. Milestones

In this project you are encouraged to use two categories of software tools available at CVC - 1. Authoring software that includes the modeling tools like TexMol, VolRover and Maya, 2. Player software that includes Perfly, Interactive VolVideo (IVV) for interactive display. For the scheduled timeline please see Table  1.


Oct 26

Project release, Discussion, Role assignments 

Nov 2

Initial Models (Molecules + Membrane)


Authoring + Player Software 

Nov 9

First unified model of NMJ  


Outline of further steps 

Nov 16

LOD (Level-of-detail) model of NMJ 

Nov 23


Nov 30

Augmenting models with properties (Electrostatics, Hydrophobicity) 


Use of 3D textures 

Dec 7

Demo Day!!! 


Table 1. Timeline


[1]  C. Bajaj and P. Djeu and V. Siddavanahalli and A. Thane. Interactive Visual Exploration of Large Flexible Multi-component Molecular Complexes. Proc. of the Annual IEEE Visualization Conference, October 2004, Austin, Texas, IEEE Computer Society Press, pp. 243-250.

[2]  C. Bajaj and V. Pascucci and D. Schikore. Fast Isocontouring for Improved Interactivity. Proceedings: ACM Siggraph/IEEE Symposium on Volume Visualization, ACM Press, (1996), San Francisco, CA. pp. 39-46 (text) 99 (color plate).

[3]  C. Bajaj and V.Pascucci and D.Schikore. The Contour Spectrum. Proceedings of the 1997 IEEE Visualization Conference, October 1997 Phoeniz, Arizona,pp.167-173.

[4]  J. R. Stiles and T. M. Bartol. Monte Carlo methods for simulating realistic synaptic microphysiology using MCell. In Computational Neuroscience: Realistic Modeling for Experimentalists ed. E. DeSchutter, pp.87-127. CRC Press, Boca Raton (2001).

[5]  Y. Zhang and C. Bajaj. Adaptive and Quality Quadrilateral/Hexahedral Meshing from Volumetric Data. In Computer Methods in Applied Mechanics and Engineering (CMAME), in press, 2005.

[6]  Y. Zhang and C. Bajaj and B-S. Sohn. 3D Finite Element Meshing from Imaging Data. In The special issue of Computer Methods in Applied Mechanics and Engineering (CMAME) on Unstructured Mesh Generation, 194(48-49), pp. 5083-5106, 2005.

4. Project Participants



Team Assignment

Description (currently empty) 

Bill Robins



Rick Bolkey



Shaolie Hossain



Albert Chen



Pai-Chi Li



Vinay Siddavanahalli



Sangmin Park



Bongjune Kwon



Jose Rivera



Zeyun Yu



Jianguang Sun



Katherine Clarridge



Fred Nugen



Powei F



Wenqi Zhao



Jessica Zhang



Samrat Goswami




Table 2. Team abbreviations: A-mol - Authoring software (Molecule modeling), A-mem - Authoring software (Membrane modeling), A-prop - Authoring software (Properties), P - Player software.

The teams and their members:

·                    Authoring Software: 

-                     Molecule team:  Bill Robins, Rick Bolkey, Vinay.

-                     Membrane team:  Shaolie Hossain, Fred Nugen, Sangmin Park, Zeyun Yu.

-                     Properties team:  Albert Chen, Wenqi Zhao, Jessica Zhang, Pai-Chi Li.

·                    Player Software:  Jianguang (Jason) Sun, Bongjune Kwon, Jose Rivera, Powei Feng.

5. Progress and Guideline

1. Membrane Group

On Friday, Nov. 4th, the group submitted the following requirement document.


Requirement document from Membrane group


Feedback from Dr. Bajaj on 5th Nov., 2005


Revised requirement document from the group on 8th Nov., 2005

2. Mol Group

On Friday, Nov. 4th, the group submitted the following requirement document.


Requirement document from Mol group


Feedback from Dr. Bajaj on 5th Nov., 2005


Response from the group on 8th Nov., 2005

3. Prop Group

On Friday, Nov. 4th, the group submitted the following requirement document.

Requirement document from Prop group


Feedback from Dr. Bajaj on 5th Nov., 2005


Revised requirement document from the group on 8th Nov., 2005

4. Player Group

On Friday, Nov. 4th, the group submitted the following requirement document.

Requirement document from Player group


Suggested IVR Data Format


Feedback from Dr. Bajaj on 5th Nov., 2005


Revised requirement document from the group on 8th Nov., 2005



6. Useful Links


The document on how to use SGE to submit a job on prisms cluster is available on prisms web page, especially there is a tutorial on how to submit APBS jobs through SGE: