You are here

distances definition in course-grained (centroid mode) steps

7 posts / 0 new
Last post
distances definition in course-grained (centroid mode) steps

I would like to define a potential between the CB atoms of a certain pair of residues in both centroid and full-atom refinement steps during Abinitio protocol. While in full-atom representation there is no ambiguity, I was wondering if the position of the CB is represented correctly in the centroid mode or this position would be the center of mass of the "super-atom" that represent the side chain, for example. If is not the first case scenario, would it be possible to calculate was it the difference between the "actual" position of CB-like atom in the centroid mode and the position it will be present in the full-atom representation? (I imagine it is a constant...).

ps. I know it would be an alternative to define in terms of CA atoms, but defining in terms of CB atoms implies an aditional accuracy to the potential.

Post Situation: 
Tue, 2017-01-03 05:15

~/ROSETTA/Rosetta/main/database/chemical/residue_type_sets/centroid/residue_types/l-caa hosts the centroid parameters files.  You will see that CB atoms are present in these parameters.  I spot checked a few and the CB position was similar to where it is in the fullatom params.  I expect CB is essentially correct; its sterics are important for ramachandran, after all.

Tue, 2017-01-03 05:30

That is good news!

Is there a place where I can access what is the meaning of each of those parameters? I mean, a description of the file... Right now, I am trying to infer them.

Tue, 2017-01-03 06:01

The params file format is older than 2007, which is a roundabout way of saying no.

  • The ATOM lines define which atoms exist (and what their chemical types are in 3 different formats, and I think partial charge)
  • UPPER_CONNECT and LOWER_CONNECT define where the connections to the next residue go (the peptide bond)
  • BOND lines define connectivity
  • ICOOR_INTERNAL lines define the geometry of connections.  The numeric parameters are a bond torsion, angle, and length defined for the connection of the atoms in the last 4 columns.  How the (possibly improper) dihedral maps to those is almost clear; I don' know how angle and bond length maps.  

I'm ignoring the other fields but can attempt to answer a specific question.


Tue, 2017-01-03 07:58

ooh, someone did write it.  It's in-code at source/src/core/chemical/ in a big comment midway though the file.


Tue, 2017-01-03 08:02

Pasting from the developer chat:

"Yeah, the docs in that file are the best description of the params file. There is also some high slightly higher level discussion in the Drew et al. NCBB paper ("

Tue, 2017-01-03 08:09

Thank you smlewis!


I will take a look. It was of a great help!

Tue, 2017-01-03 11:15