Unexpected torsions assigned for cyano group

[djcole56]

Hi,

We've been studying NNRTIs of the type described here: https://pubs.acs.org/doi/10.1021/ja408917n
and had some issues with ligpargen parameterisation of cyano group torsions.

This can be reproduced by parameterising the molecule: N#Cc2cc1ccccn1c2

I've attached the z-matrix and pdb for that compound from ligpargen. Torsion number 6 is defined as type 223. Looking at the list of opls-aa torsion types, this looks like a biphenyl-type torsion (with V2 = 2.17 kcal/mol). It looks like all cyano groups should have V1=V2=V3=V4=0 (ie type 242). The inclusion of this torsion causes instability for us during MD simulations (using the xml file). We can easily correct this by manually setting these torsion parameters to zero, but thought we should flag this up.

I'm not sure whether the CS-CS-CZ-NZ torsion type should be added to the opls-aa parameter file, or whether it's a deeper problem with ligpargen.

Thanks!

zmat.pdb.txt
zmat.txt