This is a disturbing and potentially serious bug that causes "cleanup_moltemplate.sh" to alter the order of improper atoms for all force fields, (including GAFF, GAFF2, OPLSAA, and COMPASS). It is unlikely but also possible it could have caused improper interactions to be incorrectly identified as redundant and deleted. Thanks to bhargavchava97 (again) for catching this!
2.17.1
Important correction to AMBER GAFF and GAFF2 force fields (improper interactions). Prior to this version, molecules built using AMBER force fields (GAFF and GAFF2) would contain redundant improper interactions. This might cause these molecules to be unrealistically stiff. Currently the improper interactions generated by moltemplate now agree with improper interactions generated by AmberTools/Antechamber. (Details: I can't guarantee that the order that the atoms appear in each improper is equivalent, but at least the number of improper interactions is correct, the central atom is correct, and the atoms involved are correct.) This bug fix was contributed by bhargavchava97(github user). THANKS!
2.17.0
Added the "amberfrcmod2lt.sh script for converting user-supplied FRCMOD files into moltemplate LT files that can be used with gaff.lt and gaff2.lt. I will eventually have to document all of the scripts in this directory for them to be useful to end users. (On my to-do list.) I also corrected some some of the error messages in genpoly_lt.py.
2.16.5
syntax error checking no longer crashes when it encounters {atom:b1}*{atom:b3} style ranges specifiers in pair_coeff commands (or "Pair Coeffs" sections from data files). *(Note: I have only tested input scripts with "pair_coeff" commands that use this syntax. Hopefully data files with "Pair Coeffs" also work.)*
2.16.4
The bugs effected some python environments, and also users who use atom_style ellipsoid. (Thanks to LunarLanding, Clemens Possel, and Johannes Haataja!)
2.16.3
added the "-type", "-mol", "-id" arguments to dump2data.py, to enable users to be able to select atoms based on type, id, or molecule-id. (I'm not sure if this small feature justifies a version update.)