LAMMPS Tutorial


1) Create your starting structure in SYBYL

2) If you have more than one molecule in your structure, give each of them a
   substructure name as follows:
     i) Build/Edit  >>  Modify  >>  Substructure....
     ii) Select "NAME"  >> {OK}
     iii) Select the specific substructure by clicking on it on the screen
     iv) Click {OK} on the Substructure Expression widget
     v) Type in a name (ideally just three letters)  >>  {OK}

3) If you have recently solvated your system, you can give substructure names
   to the solvent molecules more conveniently as follows:
     i) Build/Edit  >>  Modify  >>  Substructure....
     ii) Select "NAME"  >> {OK}
     iii) On the Substructure Expression widget, click {Sets...}
     iv) Click {OK} on the spurious information widget that appears
     v) Select "SOLVENT"  >>  {OK}
     vi) Click {OK} on the Substructure Expression widget
     vii) Select "SEQUENTIAL_AUTO"  >>  {OK}

4) Systematize the atomic names as follows:
     i) Build/Edit  >>  Modify  >>  Atom...
     ii) Select "NAME"  >> {OK}
     iii) Click {All}  >>  {OK}
     iv) Select "SEQUENTIAL_AUTO"  >>  {OK}

5) In general, LAMMPS will run more reliably if you do at least a partial
   minimization in SYBYL.  Depending on the size of system you plan to look
   at, you may wish to do as many as 100 SYBYL minimization iterations (for
   big systems, 20 would be a good start).  For systems with unusual atoms
   or bonding structure, you might need to use the MMFF94 force field instead
   of Tripos.

6) File  >>  Save As...  >>  choose the appropriate molecular area and type in
   an appropriate file name  >>  {Save}

7) Exit SYBYL

8) Bring up insightii

9) Molecule  >>  Get  >>  click on "Sybyl_Mol2"  >>  choose the desired file
   >>  {Execute}  >>  when it's loaded, hit {Cancel}

10) FF  >>  Select  >>  leave it as the default (cvff.frc)  >>  {Execute}

11) FF  >>  Potentials  >>  select "Fix" for both "Potential Action" and
    "Partial Chg Action"  >>  {Execute}

12) Molecule  >>  Put  >>  leave "Biosym" as a default format  >>  type in a
    name in the "Mol File Name" box  >>  {Execute}  >>  when it's done, hit

13) Session  >>  Quit  >>  {Execute}

14) rsh onto the cluster and copy the *.car and *.mdf files created in step 11)
    over to a directory created for this calculation

15) Make sure that the execution directory also contains a copy of the cvff.frc
    force field parameters, and a LAMMPS input file.  Copies of these two files
    may be obtained on the cluster at:  /home/ghl/bridget/lammps_inputs

16) Inside that directory, run the following command:
      msi2lmp sample >
    where you should substitute for "sample" the prefix of your *.car and *.mdf
    file names.  This will create a LAMMPS data file with a name like

17) Edit your LAMMPS input file file to ensure that the file names specified
    match the file names you're using.  I.e., the "read data" line should give
    a file name like "".  Also modify the file names specified in
    the "dump atoms" and "restart" lines

18) Submit your LAMMPS job:
      i) Type the command:  lmps2
      ii) Provide the full name of your LAMMPS input file you edited in 16)
      iii) Provide the full name of your LAMMPS data file created in 15)
      iv) Specify a number of processors; if the script gives you an error
          message, then try a smaller number.

19) The job will run some amount of time (anywhere from a few seconds to days).
    Check the output file (*.out) to observe its progress.

20) When your job has completed, you can generate coordinates that you can view
    using SYBYL as follows:
      i) Type the command: coorout
      ii) Give the name of your LAMMPS data file
      iii) Give the name of the coordinate dump file (*.dump) that you want to
           look at
      iv) Copy over the resulting lmpcrd.mol2 file to an SGI machine to view,
          and be careful with it because some atom types may get changed during
          the conversion.

David Johnson
Access to more than 40 computational chemistry software programs and databases
High-performance computational tools accelerate drug discovery and minimize costs
Analyze complex, multidimensional data sets to rapidly generate biological insights
KU Today