BOSS Pure Liquid Simulation Notes

From Rizzo_Lab
Revision as of 13:56, 27 February 2008 by Rizzo (talk | contribs)
(diff) ←Older revision | view current revision (diff) | Newer revision→ (diff)
Jump to: navigation, search

Simulations of a Pure Liquid : AMS-536 Molecular Modeling of Biological Molecules

Our first calculation will be for a pure liquid using the BOSS program from the Jorgensen group at Yale (http://zarbi.chem.yale.edu). The first step is to download and read the users manual for BOSS from the Jorgensen Group website. The BOSS manual takes you through the parameter input file, defines the various parameters, goes through a sample zmatrix line-by-line, discusses various types of calculations, and defines variable definitions and portions of the output.


Several files needed to run the calculations are in my home directory on the Mathlab computers: /home/rizzo/AMS536projects/pureliquid. We will also discuss the details in class. Under “pureliquid” are 3 directories: MCgasopt, MCgasruns, and MCliqruns which contain various files and scripts needed for the calculations. I would suggest copying over the contents of my pureliquid directory into a directory named “AMS536projects” in your home accounts on both the MATHLAB computers in the SINC site and also on the SEAWULF supercomputer. Note to copy files over to SEAWULF make a tar file (tar -cvf pureliquid.tar pureliquid )and then use the sftp protocol to copy the file “pureliquid.tar” the SEAWULF. Untar the file using: tar -xvf pureliquid.tar


Note, we are actually running boss test job #7 (see the BOSS manual pages 40 and 52-54). I simply made a few cosmetic changes to the "cmd" scripts to make things a bit more clear for the class. Note that since you will be editing the various cmd scripts make sure your “vi” skills are up-to-date. FYI all the boss test jobs discussed in the manual are included in our boss distribution so feel free to explore.


Make copies from /home/rizzo/AMS536projects/pureliquid to your directory and edit the $MYdir parameter in the files OPTcmd, MCliqcmd, and MCgascmd to specify the directory on your account in which BOSS will be reading and writing files.


If you are running very short jobs to test things in the Mathlab then you can execute scripts by typing (for example) at the prompt: csh OPTcmd >&log& If you want to execute jobs on SEAWULF you must submit the jobs using a queing system called “PBS”. See http://seawulf.stonybrook.edu/ (follow links to support and FAQ). Basically, for a “serial job” i.e. one processor you can just type as an example: qsub OPTcmd


To see which resources are available (i.e. free nodes to use) go to the SEAWULF webpage http://seawulf.stonybrook.edu/ganglia and choose the source to be “nodes” and then sort the nodes by hostname. Nodes without any jobs running show up in light blue.


The first calculation to run is in directory “MCgasopt. Copy over a zmatrix file for a molecule from the boss distribution “aazmat” directory to the MCgasopt directory and edit the parameter $molecule in OPTmcd to be the name of the zmatrix you just copied over. Run the shell script csh OPTcmd >&log& and when complete look at all the various input and output files to understand what they are used for. When in doubt consult the BOSS users manual. Look at the molecule visually and optimize a variety of molecules in the aazmat directory. Note that the optimized zmatrix from running OPTcmd is called “sum”. Open up sum with “vi” and look at it. The “plt” files are in PDB format and can visualized using a number of programs such as rasmol, vmd, or moe. Note that the “sum” file is a fully expanded zmatrix file which we will use for the other two BOSS calculations in directories MCgasruns and MCliqruns to compute pureliquid properties.


After successfully executing OPTcmd and understanding all input and output then run the next two simulations in MCgasruns and MCliqruns. Make any the necessary changes (i.e path names for where the files will be read and written) to the scripts MCgascmd and MCliqcmd and edit the parameter files so that the temperature is correct for the experimental data for which you want to compare. For example most jobs will be run either at 25 deg C or at the normal boiling point of the pure liquid. These jobs take some time so execute MCgascmd and MCliqcmd on Seawulf.


Everyone is assigned a different liquid to simulate. See the INDEX file in the "aazmat" directory which contains a list of zmatrices for the common organic molecules we will simulate.


Run a pure liquid simulation and a gas-phase simulation for your molecule at 25 C if the molecule is a liquid at 25 C. If not, then your simulation should be run at the normal boiling point of the liquid. Use Tables 9-10 from Jorgensen et al. J. Am. Chem. Soc. 1996, 118, 11225-11236 (Jorgensen002.pdf) to get the simulation temperature and the experimental density for which you will compare your computational results for your molecule(s) assigned in class.


To analyze your results I suggest the following:


  1. Look at a representative plt file using one of the visualization programs in the lab.
  2. Plot the total system energy versus number of Monte Carlo steps. Note BOSS executes jobs in "blocks" with each block having of a defined number of "configurations" (i.e. MC steps). Look at the "cmd" scripts if this is unclear.
  3. Compute the density of the pure liquid in g/cm^3
  4. Compute the heat of vaporization in kcal/mol


It is important to understand what the scripts are doing and what the input and output files contain. The purpose is not to just run the calculations but understand the process by which the calculations are actually setup so you can modify things in the future.


If you are stuck send email to the whole group. Everyone please reply to these group questions so things don't get out of hand.