VMD-L Mailing List

From: Bo Højen Justensen (04bhj_at_nanostud.aau.dk)
Date: Tue Mar 04 2008 - 02:33:11 CST

Hi Stephen,
Actually, I start by calculating the area of all the voronoi cells in a snapshot. I then check if they actually have an area. If no area is returned it means that the voronoi cell has a vertex located in infinite and the point/lipid is on the boundary. This dosent catch all the boundary lipids though, so at the same time I also set a threshold area, above which the lipid is believed to be on the boundary. This method cant be used of course, if one dosent have a voronoi tessellation of the points at all.
I havnt got any experiense using PBC and Mathematica, but doing a voronoi tessellation on a sample of points located in 2D space is straightforward in matlab.

Best regards,
BHJ.

________________________________________
Fra: Stephen Dutz [smdutz_at_csupomona.edu]
Sendt: 3. marts 2008 23:06
Til: Bo Højen Justensen
Cc: vmd-l_at_ks.uiuc.edu
Emne: RE: average area per lipid

Hi Bo,

Thanks for the response! How do you go about excluding the boundary lipid? Do you exclude based on distance from the center of the box? Everything is failing for me at the tessellation step, which means I would have to exclude the boundary lipid prior to that.

Thanks again,
Stephen

----------
"You must be the change you want to see in the world."
-Mahatma Gandhi

________________________________

From: Bo Højen Justensen [mailto:04bhj_at_nanostud.aau.dk]
Sent: Mon 3/3/2008 1:08 PM
To: Stephen Dutz
Subject: SV: average area per lipid

Hi Stephen,
I've calculated the average area per lipid using the same method as you, that is a Voronoi tesselation of coordinates from a representative atom of the lipids. I use Matlab to do it, and have written a small script that calculates the area per lipid for each snapshot and then average it over the total simulation. To rule out border effects I exclude the outer lipids and lipids with an area above a certain threshold which indicate that they are on the outer border and not representative of the bilayer. This method works fine, and the only problem I have is me not being able to do an Voronoi tesselation on an 3D surface, limiting the calculating of the area per lipid to a 2D projection of the coordinates. This is of course not optimal when you have a lipid surface that is "coiling and twisting". I dont know if you can use this information to anything but a prove that it can be done, using matlab anyway.

Best regards,
Bo H. Justesen
________________________________________ T
Fra: owner-vmd-l_at_ks.uiuc.edu [owner-vmd-l_at_ks.uiuc.edu] På vegne af Stephen Dutz [smdutz_at_csupomona.edu]
Sendt: 3. marts 2008 18:14
Til: vmd-l_at_ks.uiuc.edu
Emne: vmd-l: average area per lipid

Hi all,

I'm using gromacs for MD simulations on membrane proteins. As part of the analysis I would like to calculate the average area per lipid. I've been going about this by pulling the coordinates from the trajectory in VMD, then importing these to Mathematica and using it to plot a voronoi tessellation. The problem I run in to is I have to discard ~10% of the data (when averaging over multiple frames) due to Mathematica not being able to complete the tessellation. I believe this is occurring as a result of PBC and Mathematica plotting concave polygons. That being said, I have just about given up on my method.

I was wondering if anyone out there has a script or methodology they would be willing to share with me in order to calculate the area per lipid with a protein present (i.e using a voronoi tessellation). I would appreciate it tremendously!

Sincerely,

Stephen Dutz

----------
"You must be the change you want to see in the world."
-Mahatma Gandhi