This forum attends all concerns of the Java-based MembraneEditor created for the CELLmicrocosmos 2 project.
Post Reply
User avatar
bjoern
Site Admin
Posts: 211
Joined: 20.03.2006, 15:01
Contact:

Questions

Post by bjoern » 07.07.2006, 09:38

Here you can place questions concerning the Cm2 MembraneEditor.

Philipp
Posts: 7
Joined: 30.07.2014, 09:57

Re: Questions

Post by Philipp » 30.07.2014, 11:01

Hi, some things cross my mind:
  • What about more complex formes of membranes like fusion events of lipsomes and membranes?
  • Would it be possible to import a base mesh as template for placing lipids?
  • Maybe a warning would be nice if you enter unreasonable small values for the semi-axis in the Vesicle Builder because you can not stop the initial placing progress as discribed in the bug report.
  • How about giving a fix number of lipids instead of fixing the vesicle size?
Best,
Philipp

User avatar
bjoern
Site Admin
Posts: 211
Joined: 20.03.2006, 15:01
Contact:

Re: Questions

Post by bjoern » 30.07.2014, 13:00

What about more complex formes of membranes like fusion events of lipsomes and membranes?

This is already possible. Just use the function "Add secondary bilayer to the model". The button you will find in the toolbar. With this option, you can add as many membranes or bilayers as you wish. And the collision detection works for all membranes. Therefore, if you place e.g. a liposome on top of a membrane, only lipids which do not intersect will be placed. And of course you can afterwards manually delete all unnecessary lipids.

Would it be possible to import a base mesh as template for placing lipids?

Basically, this would be possible, but it is not implemented yet. We have some similar code used for another project here, but it was not used for the MembraneEditor by now. So, this is a good idea for a new Membrane Packing Algorithm!

Maybe a warning would be nice if you enter unreasonable small values for the semi-axis in the Vesicle Builder because you can not stop the initial placing progress as discribed in the bug report.

Please see my answer here:
https://www.cellmicrocosmos.org/Cmforum ... 1533#p1533

How about giving a fix number of lipids instead of fixing the vesicle size?

This is a good idea for a future version of the Vesicle Builder!

Philipp
Posts: 7
Joined: 30.07.2014, 09:57

Re: Questions

Post by Philipp » 30.07.2014, 14:00

Thanks for the quick reply. I was hoping for this project to be still alive :)
This is already possible. Just use the function "Add secondary bilayer to the model". The button you will find in the toolbar. With this option, you can add as many membranes or bilayers as you wish. And the collision detection works for all membranes. Therefore, if you place e.g. a liposome on top of a membrane, only lipids which do not intersect will be placed. And of course you can afterwards manually delete all unnecessary lipids.
I tried this approach, but it is not what I meant. The membrane and the vesicle are still independant objects, hence the highly curved section where the vesicle fuses with the membrane is lacking. That's why I asked for the possibility to import an externally generated template.

User avatar
bjoern
Site Admin
Posts: 211
Joined: 20.03.2006, 15:01
Contact:

Re: Questions

Post by bjoern » 30.07.2014, 19:21

Of course, this project is very lively - although the discussions in the forum could be more! "alive", yes :mrgreen: So thanks to your revival!

And I have a lot of stuff to extend the MembraneEditor lying around here which is waiting for publication but I did not find the time by now - too much teaching these days, but we are working on it ;-)

Okay, so I see your problem. I have an extended version of the Wanderer lying around here where I have to fix a small issue, then you could to similar things like you say, but you will not be able to model kind of droplet. But you could model membranes with different curvatures by using a function of the third order. But: to use a mesh, this will be more work.

Philipp
Posts: 7
Joined: 30.07.2014, 09:57

Re: Questions

Post by Philipp » 10.09.2014, 13:59

Hmmm ... maybe there is some kind of .stl import plugin you just have to copy and paste. But I don't have an coding experience with Java, so I don't know the pitfalls.

I am again playing around with this membrane fusing stuff (see the bug report ^^) and I want to ask if it is possible that the Advanced Random Placing algorithm remembres the overall density of lipids? Since you can not enter a value the domians you are working on could end up having different densities. Maybe the slider can maintain its last position.

If you alter the dimnesions of your membrane, would it be possible that the box for the Z-value is automatically marked (like you used a mouse double click) if you switch from the X-value to the Z-Value using TAB?

Would it be possible to enlarge the window with the forum? Scrolling around in X-direction because the entries are wider than the window is somehow annoying.

Best,
Philipp

\\edit
Maybe you could prolong the time out limit after which you get automatically logged out? If I did not copy the bug report before sending it, it would habve been gone since I was logged out :/

User avatar
bjoern
Site Admin
Posts: 211
Joined: 20.03.2006, 15:01
Contact:

Re: Questions

Post by bjoern » 10.09.2014, 15:27

Philipp wrote: Hmmm ... maybe there is some kind of .stl import plugin you just have to copy and paste. But I don't have an coding experience with Java, so I don't know the pitfalls.

[...]

Maybe you could prolong the time out limit after which you get automatically logged out? If I did not copy the bug report before sending it, it would habve been gone since I was logged out :/
Hi Philipp!

Thanks for you comments and the bug report! I see, you have some jobs for me to do, okay, I hope I will find some time to solve at least some of your questions in the next weeks. And thanks for the bug report, so I will try to solve some of the issues for the next release.

But some first answers:

The STL format is a binary format. I have never used it, but I think you talk about this here:

http://en.wikipedia.org/wiki/STL_%28file_format%29

So it will be not so easy, because these files will usually not contain any information about the atoms - only some coordinates representing positions and shapes. So I fear, you cannot just create a PDB file from this STL file by writing a script. I see that they are often used for 3D printers and that tools like Chimera are supporting this. Maybe you can do something with Chimera, but I think, it will be only for export. Just test it. So usually, the easy way would be to find out which molecule is shown, and then just try to find it at one of the public PDB databases. I listed many here:

https://www.cellmicrocosmos.org/Cmforum ... p?f=9&t=20

But one question above I may solve already completely! Concerning the annoying scrolling: there is a hint on the top of the forum page, it looks like this:

forum.CELLmicrocosmos.org

Maybe, you should just try what your browser will do with this link, and one problem might be solved. Usually, the content of the forum page should fit the frame of the main webpage, but if you use your mobile phone or a small resolution you might run into problems with the space, yes!

About the time out, I looked it up in the configuration, and you should have 2 hours time - this was not long enough? Or the configuration of the forum is somewhere ignored. So after how much time have you been logged out?

One more question: why do you not just use the distributor which is optimized for defining lipid densities? In the background, the distributor uses the same random placing method like the Adv. Rand. Placer. But it is not completely the same method, because Adv. Rand. Placer usually subdivides the membrane into smaller sub areas during the placing process. Still, at the moment you have to copy/paste then the density values if you use different domains or layers.

Cheers,
Björn

Philipp
Posts: 7
Joined: 30.07.2014, 09:57

Re: Questions

Post by Philipp » 10.09.2014, 22:58

bjoern wrote: So it will be not so easy, because these files will usually not contain any information about the atoms - only some coordinates representing positions and shapes. So I fear, you cannot just create a PDB file from this STL file by writing a script.
Sure and that is not what I want to do. Does the grid in your editor contain any information about atoms before adding any lipids? I thought this is just a 2D-bounding box for placing lipids and proteins. That's why I came up with the .stl-file import question. I thought of replacing your grid with a user defined grid. Because atoms follow the grid, the grid itself does not need any informations in the first place, right? But I don't know how to take care of the right propotions. Of course, I could prepare a grid that matches a 200x200 A grid of your editor plus a dimension in Y direction (that would be the fusion vesicle ;) ) and I could make as dense a you need it. But as stated before, I don't know how complex the implementation would be.

What for? At the moment my workflow looks like this: I create a membrane with your tool and export it as .pdb. I open the file in VMD to create selections according to lipid composition and to separate headgroups and tails. Then I export the selections as .stl-file and import these in Blender for final tweaking and rendering. I know that Blender can handle .pdb-files, but just on a very very basic level. But in Blender I can not adjust the amount of lipids if I start to bend the membrane. Moreover the lipids get distorted. Its all about the curvature ;)

Using .stl-files out of VMD in Blender is the reason I am trying to set up a triple layer system where I sandwich a vesicle in-between two normal membranes. Doing so I can get a lipid ring instead of a complete vesicle to construct a fusion event in Blender. Hopefully this will decrease the size of the .stl-files I have to import in Blender. Example: A vesicle with 69 A semi-axes (smaller diameter produces artefacts) results in a 8 GB .stl-file :?
Today this triple layer system did not work. In the first attempt I got stuck in the inital adding phase without the possibility to interrupt the process and in the second try I could not start the Vesicle Builder because the default layer was said to contain no lipids (not true).

Maybe I am misusing your program, but it is fun playing around :mrgreen:
bjoern wrote: But one question above I may solve already completely! Concerning the annoying scrolling: there is a hint on the top of the forum page, it looks like this:

forum.CELLmicrocosmos.org

Maybe, you should just try what your browser will do with this link, and one problem might be solved. Usually, the content of the forum page should fit the frame of the main webpage, but if you use your mobile phone or a small resolution you might run into problems with the space, yes!
That's the solution. Up to now I came from your main page to the forum. And doing so, the forum is displayed in this small inset in your main page. I used a normal PC.

bjoern wrote:About the time out, I looked it up in the configuration, and you should have 2 hours time - this was not long enough? Or the configuration of the forum is somewhere ignored. So after how much time have you been logged out?
Hm ... I have no idea. I was just writing the report. It took its time, but two hours ... Next time I use a stop watch :lol:
bjoern wrote:One more question: why do you not just use the distributor which is optimized for defining lipid densities? In the background, the distributor uses the same random placing method like the Adv. Rand. Placer. But it is not completely the same method, because Adv. Rand. Placer usually subdivides the membrane into smaller sub areas during the placing process. Still, at the moment you have to copy/paste then the density values if you use different domains or layers.
Ah ... you got me there. As a good scientist I took a quick, not so deep look in the description of the algorithms. The Wanderer Rookie works fine and the resulting membrane looks nice. So I stuck with it. The Adv. Rand. Placer is the first algorithm listed and therefor the first choise for the domains. I will have a closer look into the documentation :mrgreen:

Best,
Philipp

Philipp
Posts: 7
Joined: 30.07.2014, 09:57

Re: Questions

Post by Philipp » 28.11.2014, 10:19

Hi, its me again :wink:

Would it be possible to get upper and lower leaflet as individual chains in the generated .pdb-file? Of course I you could separately generate both layers but that would double the time. I am asking this because there is always a defined space between both layers and since I don't have neither the time nor enough CPU/GPU power to run a simulation. I would like to close the gap manually either in VMD or in Blender.

Best,
Philipp

User avatar
bjoern
Site Admin
Posts: 211
Joined: 20.03.2006, 15:01
Contact:

Re: Questions

Post by bjoern » 12.01.2015, 13:54

Philipp wrote:Hi, its me again :wink:

Would it be possible to get upper and lower leaflet as individual chains in the generated .pdb-file? Of course I you could separately generate both layers but that would double the time. I am asking this because there is always a defined space between both layers and since I don't have neither the time nor enough CPU/GPU power to run a simulation. I would like to close the gap manually either in VMD or in Blender.

Best,
Philipp
Hi Philipp,

sorry, I did not see this post! I think, for your purpose you can just use the MembraneShifter tool. With this tool, you can just move all lipids of the upper leaflet or only those ones of the bottom leaflet into one direction by changing the Y shift value. In this way, you can change the distance between both layers directly.

By the way, did you check out the new version of the Vesicle Builder? I improved it a lot and fixed many bugs:
http://www.cellmicrocosmos.org/Cmforum/ ... f=35&t=676

Also, your sandwich model should be no problem.
1. Generate a bilayer.
2. Add a new bilayer, at Y=100.
3. Generate a vesicle at Y=50 (in the center of the two bilayers).
4. During PDB export, say "Yes" if it is asked if you want to export the whole membrane.

So I just tried it and it worked.

User avatar
bjoern
Site Admin
Posts: 211
Joined: 20.03.2006, 15:01
Contact:

Re: Questions

Post by bjoern » 12.01.2015, 13:59

Philipp wrote:
bjoern wrote: So it will be not so easy, because these files will usually not contain any information about the atoms - only some coordinates representing positions and shapes. So I fear, you cannot just create a PDB file from this STL file by writing a script.
Sure and that is not what I want to do. Does the grid in your editor contain any information about atoms before adding any lipids? I thought this is just a 2D-bounding box for placing lipids and proteins. That's why I came up with the .stl-file import question. I thought of replacing your grid with a user defined grid. Because atoms follow the grid, the grid itself does not need any informations in the first place, right? But I don't know how to take care of the right propotions. Of course, I could prepare a grid that matches a 200x200 A grid of your editor plus a dimension in Y direction (that would be the fusion vesicle ;) ) and I could make as dense a you need it. But as stated before, I don't know how complex the implementation would be.
About the STL file: we are working on a tool which will do what you are asking for (together with Prof. Mario Botsch and Thomas Waltemate). Because of the complex meshes, this is not a task for the MembraneEditor but it will be compatible to the MembraneEditor.

The Alpha-Version of this tool is discussed here:
http://diglib.eg.org/handle/10.2312/vcb ... 87.089-096

Currently, a student is working on a version which will suport PDB export. I will let you know, if there is a download available. So input will be some 3D file and a PDB membrane, and export will be the PDB membrane in a shape surrounding the mesh file.

Post Reply