Artoria2e5 Lv 1
On the rosetta side they've change the script up a bit, at https://www.rosettacommons.org/docs/latest/application_documentation/structure_prediction/relax#description-of-algorithm. Changes are:
- CI order.
- Now they have a new
coord_cst_weight, which seems to be some sort of band-in-space to discourage super huge movements.
The documentation also mentions a whole family of relax scripts. The different CI schedule can be found in PMC8299543 supplementary info table SI1, https://europepmc.org/articles/PMC8299543/bin/NIHMS1650092-supplement-supinfo.pdf. PMC8299543 / PMID 33249652 can be treated as the primary citation for this new family of relax functions with different shake and wiggle CI.
I think I can find enough time to get a new version out, minus the BiS part.
Another discrepancy is minimization tolerance. I have no idea how to even approximate that one. Perhaps with "Rosetta energy" calculation and cycle-by-cycle score change tracking?