[Swift-commit] r5902 - SwiftApps/CMTS/dimers

[wilde at ci.uchicago.edu]

Author: wilde
Date: 2012-08-13 11:47:42 -0500 (Mon, 13 Aug 2012)
New Revision: 5902

Added:
   SwiftApps/CMTS/dimers/README_USE_CASE
Log:
Add description of this use case from Johns email

Added: SwiftApps/CMTS/dimers/README_USE_CASE
===================================================================
--- SwiftApps/CMTS/dimers/README_USE_CASE	                        (rev 0)
+++ SwiftApps/CMTS/dimers/README_USE_CASE	2012-08-13 16:47:42 UTC (rev 5902)
@@ -0,0 +1,103 @@
+From John Grime, 7 Aug 2012:
+
+---
+
+Chaps,
+
+Please find attached a compressed directory with an example of the
+sort of thing I sometimes need to do which may benefit from
+SWIFTification.
+
+---------------
+Brief overview:
+---------------
+
+I have some molecular simulations where I try to get a set of proteins
+to self-assemble into a larger structure inside a spherical region of
+space.
+
+I know what the "target" concentration of protein is for the
+biological system I'm trying to model, but I found that when I add all
+the proteins at once I get multiple "nucleation sites" - instead of a
+single structure forming, I instead got multiple incomplete
+structures.
+
+To get around this I instead tried to add the proteins gradually, a
+few at a time. The general workflow is therefore something like this:
+
+1. Add a few proteins to the simulation data.
+
+2. Run the molecular dynamics simulation for a while, save a snapshot
+of the simulation at the end of the run.
+
+3. Take the snapshot written at the end of step 2, and use as the
+input for step 1.
+
+-----------------------
+What I need to do this:
+-----------------------
+
+1. The molecular dynamics program itself; I use a modified version of
+   LAMMPS, but in principle I could probably use other codes.
+
+2. A little program to insert a specified number of molecules into an
+   existing simulation, given a well-defined molecule description.
+
+3. The ability to convert a binary LAMMPS snapshot into something more
+   convenient for the previous program to deal with.
+
+--------------------------------------
+Description of files in the directory:
+--------------------------------------
+
+lammps_new_spherical.in :
+
+A LAMMPS control file, containing miscellaneous information
+(temperature, interaction parameters etc) to describe how to run the
+simulation.
+
+dimer_1232.lammps_config :
+
+A LAMMPS format molecular description of the protein we're adding to
+the simulation (atom types and coordinates, what's bonded to what etc)
+
+empty.lammps_config :
+
+The initially empty simulation that we start to add proteins into.
+
+GeometryUtil.*, Ran1.*, StringUtil.*, definitions.* :
+
+Utility code!
+
+insert_molecules.cpp :
+
+The program to insert molecules into an existing simulation
+definition. Assumes LAMMPS config format, such as found in
+empty.lammps_config and dimer_1232.lammps_config.
+
+restart2data.cpp : 
+
+A program to convert the binary LAMMPS snapshot format into something
+suitable for use with the "insert_molecules" program.
+
+---------------
+Notes
+---------------
+
+Rather than include the source code for the modified LAMMPS MD
+program, and leave you guys to it, I've instead included the script
+which runs the little protein insertion loop and commented out the
+line which runs LAMMPS. I have, however, included some example binary
+output files from LAMMPS which should allow the loop to work as if we
+actually were running LAMMPS in the loop.
+
+Hopefully, all should be pretty straightforward if you look at the
+"looped_run.sh" script. The "insert_molecules" and "restart2data"
+programs are compiled when you run the script if the binaries don't
+exist.
+
+The version of this script I actually use has some PBS stuff in there
+to generate a set of submission scripts which wait on previous job
+completion, but I thought I'd keep it simple in this case!
+
+J.