moltemplate prefers pair_coeffs over "Pair Coeffs"

I proceded to copy the pair coefficients defined in the oplsaa.lt fie.
  write_once("In Settings"){
    pair_coeff @atom:80 @atom:80 lj/cut/coul/long 0.066 3.5
    pair_coeff @atom:85 @atom:85 lj/cut/coul/long 0.03 2.5
  } #(end of pair coeffs)

I then proceded to edit the Data Pair coefficients into the following and
then pasted it into the system.lt file as follows:
write_once("Data Pair Coeffs"){
        pair_coeff @atom:80 lj/cut/coul/long 0.066 3.5
         pair_coeff @atom:85 lj/cut/coul/long 0.03 2.5
} #(end of pair coeffs)

   To get past this particular problem, you must also delete the
"pair_coeff" at the beginning of each line in "Pair Coeffs", but let's
back up a bit:

   These days, many data files contain "Coeff" sections (force-field
parameters). They are automatically included in the data files
created by the write_data command in LAMMPS.

   However, for the sake of robustness and generality, programs like
moltemplate.sh and ltemplify.py try to encourage users to use input
script commands instead. In fact, the "ltemplify.py" file converter
actually does this automatically. (It attempts to convert all "Coeff"
sections found in a data file into the corresponding input-script
commands. Given that the syntax could be totally different, I suppose
I should make this feature optional. -In a future release I'll do
that.)

    There are two potential problems:
  1) Some pair_styles (especially the tabular and many-body
pair_styles) have so many parameters that they can not be stored in a
single line of text. If my understanding is correct, you currently
cannot store these kinds of parameters in a data file.

  2) Also some of the pair_styles have buggy read/write_data() or
read/write_restart() method functions. (I introduced one of these
bugs myself!) Additionally, a few years ago I noticed a bug in
angle_harmonic.cpp in it's read/write_restart(). (It has since been
fixed.)

   The advantage to using pair_coeff commands (and bond_coeff,
angle_coeff, dihedral_coeff, improper_coeff commands), is that you can
use them to override any inaccurate force-field parameters that may
have been read from a data file or a restart file.

    For common styles, you don't have to worry about any of these
issues (eg AMBER, or OPLS force-field compatible styles like
pair_style lj/cut/coul/long, angle_style harmonic, dihedral_style
opls). In this case, it's probably fine to leave those parameters in
the the data file. But you have to do it consistently. You won't be
able to mix these LT files with others which do it the other way
(using input script commands).

   See below for how to do that.

Andrew

P.S.
    If you really want to manually move the input script commands into
the data file, you must delete all of the "pair_coeff", "bond_coeff",
"angle_coeff", "dihedral_coeff", "improper_coeff" commands and copy
the text in the corresponding section of the data file. (And you must
careful to delete these "coeff" commands from the beginning of each
line. You must also delete the extra atom-type in each "pair_coeff"
command, since the "Pair Coeff" section only allows you to specify one
atom type. Class2 force-fields require special attention, because
they have non-standard data-file section names, such as "Angle Angle
Coeffs", "BondBond13 Coeffs" etc...)

   But there are other problems with your files which caused the
original error. I'll email about those separately.

[...]

   To get past this particular problem, you must also delete the
"pair_coeff" at the beginning of each line in "Pair Coeffs", but let's
back up a bit:

   These days, many data files contain "Coeff" sections (force-field
parameters). They are automatically included in the data files
created by the write_data command in LAMMPS.

FYI, as of recent, the write_data command has a "nocoeff" flag that
tells it to skip over writing out *Coeff sections.

axel.