Dear developers,
Does the new mechanism of secondary structure restraints apply to nucleic acids as well ? Are the average bond distances similar and does DSSP works well with RNA/DNA ?
No. The DSSP algorithm only deals with proteins, and the hydrogen bonding patterns in proteins are much easier to describe using reduced representations (3 types of helix, 2 types of sheet) than for nucleic acids. As always, you can specify hydrogen bonds individually using the custom bond definitions in phenix.refine:
Jeff Headd started working on extracting H-bond information from the output of Probe, but I think we were still trying to figure out a more elegant way of storing these. Specifying helices with Watson-Crick bonding alone is going to miss many of the interactions in large RNAs, so listing individual bonds may be the best option. It would probably not be difficult to add a command to generate the list automatically (e.g. "phenix.base_pair_restraints"). We are working on ways to make these bonds easier to manage and visualize, and Bradley Hintze (Richardson lab, formerly Sean Johnson's lab) has written a PyMOL plugin to pick and display individual bonds and generate phenix.refine restraints (which I can't find online, but I can email him if you're interested).
I have no idea what the appropriate distances are in nucleic acids - I imagine they're similar for genuine H-bonds (~1.975A), but probably more varied if you try to approximate them without hydrogens.
-Nat