Merge pull request #2293 from RosettaCommons/jadolfbr/glycan_data_updates
Jadolfbr/glycan data updates
This PR updates the default glycan conformer sampling table with new data derived from the PDB using Adaptive Kernal Density estimates (same methods as the 2010 rotamer library) and a Von Misses Kernal to derive dihedral bins for each glycan linkage type seen in the data.
Raw Data was provided by Thomas Lutteke of Glycosciences.de and the densities were generated by Maxim Shapovolov and Roland Dunbrack.
This PR also updates a few default parameters to Glycan Relax and adds a few options for benchmarking. Integration test changes expected to tests that use Glycan Relax.
Density estimates were used to generate cubic spline(s) in matlab in order to identify torsion bins and assign conformers to each linkage data point using the inflection points of the spline to define single-dihedral conformer bins.
For example, the Chi1/Omega2 of an ASN a-D-GlcpNAc has three bins, as identified by the maxima of the density and boundaries at the minima on either side of each maxima.