I am trying to use allosmod package in MODELLER for ab initio modeling of glycans as described in the paper "All-atom ensemble modeling to analyze small-angle x-ray scattering of glycosylated proteins" (Structures 2013. DOI: 10.1016/j.str.2013.02.004).

I have two questions regarding the glycan modeling method, about which I could not find any prior discussions in this forum. My questions are:

(1) In the glycan modeling methods section of the paper, it is mentioned that "initial structures were generated by addition of glycan chains with ideal geometries derived from CHARMM, followed by a 1angstrom randomization of the full atomic coordinates". I am assuming this randomization is the same as the randomize_xyz function of MODELLER. However, since templates are not being provided by the user for the glycans, how are the sugar ring puckerings and dihedral angles correctly reproduced even after such large random deviations?

(2) My understanding is, in MODELLER, the solvation information is taken implicitly from the template provided. However, since the glycans are modeled template-free, is there any explicit treatment of solvation, or is the optimization done in vacuum?

Thanks,

-Srirupa