You are here

about Residue Number in Rosetta

8 posts / 0 new
Last post
about Residue Number in Rosetta

Does Rosetta use the residue number written in the pdb files? Or it just take the first shown residue has the residue number 1?
I need to clarify this issue so that the constrains are applied correctly.

Thank you very much.

Post Situation: 
Mon, 2012-06-11 14:58

The answer to the general question is "it depends on where the numbering is being used".

Generally, whenever a chain letter is specified along with the residue number, the numbering corresponds to the PDB numbering. Whenever a residue number is specified as just a single number (without an alphabetic chain designator) the numbering is "pose numbering" - that is, each residue numbered sequentially starting at 1. (Be careful, as occasionally residues in the input PDB aren't loaded into the pose, and as such get skipped in pose numbering.)

For the standard constraint file format (, the numbering is pose (sequential starting from 1) numbering, consistent with the fact that there isn't an opportunity to specify the chain.

(Note that there's been discussion about supporting PDB numbering in constraint files, but I don't believe that's been developed enough to be in/be usable with any of the released versions.)

Mon, 2012-06-11 17:23

Generally speaking:

Resfiles: PDB numbering (includes chain; supports insertion codes)

Loops files: "pose numbering" as Rocco designated

Constraints files: pose numbering

Anything else: Pose numbering unless it also requests a chain, in which case it's probably PDB numbering.

I have a Q&D app to strip PDB info and turn it into a Rosetta-fied PDB called "ForgetPDBInfo" if you need it. I think there may also be an option you can pass to the score or score_jd2 app that does the same thing.

Thanks for asking, by the way - we get a LOT of questions here that are caused by people using PDB numbering where they should have used pose numbering.

Mon, 2012-06-11 18:00

Really appreciate your answers. I think I need this app, where can I find it?

Thank you very much!

Tue, 2012-06-12 08:23

Try using score_jd2 with -renumber_pdb first.

Tue, 2012-06-12 08:45

Hey there,
wondering about the same problem I found that at least for SiteConstraint a chain ID and a residue number are given in the Docking manual of 3.4 under Tips:

Does that mean this residue number is the one in the pdb file then in the respective chain?

Also for the other constraint types residue numbers with chain identifiers are given in the RosettaCon2011Demo "dock_with_user_defined_constraints"...

Any clarification would be highly appreciated :)

Additionally, when trying to use AmbiguousConstraint Constraint1 ... Constraint2 ...
only Constraint1 is mentioned in the output/logfile. AmbiguousConstraint is not mentioned in the output/logfile.
Is that normal or indicative of a mistaken I may have made?

Wed, 2012-06-20 07:44

For constraints, you should be able to specify residues either way (at least for recent Rosetta releases). If you specify it with a postfixed chain letter, it uses PDB numbering (the number of the residue in the PDB file), whereas if you specify it as a bare number without a chain, you would use pose numbering (sequential numbers starting at 1). (Note that you currently can't use PDB numbering with residues with insertion codes - you would have to use pose numbering for them.)

The constraint input code is sometimes rather brittle. I think the issue you're running into with the AmbiguousConstraint is that the AtomPairConstraint (as well as SiteConstraint and probably a number of others) silently eats the rest of the line.
The "approved" way of dealing with AmbiguousConstraints and other types of multiconstraints is to put each subconstraint on its own line:

CoordinateConstraint OD1 6 CA 6 95.723 49.348 9.574 BOUNDED 0 0.1 0.5 0.5 tag
CoordinateConstraint ND2 6 CA 6 95.723 49.348 9.574 BOUNDED 0 0.1 0.5 0.5 tag

(Note that as the residue number is not followed by a chain, it's being given with pose numbering.)

Wed, 2012-06-20 11:37

BOUNDED in particular requires rswitch and tag, in contradiction to earlier documentation. It will silently eat through the next line of constraint code otherwise.

Sun, 2012-06-24 13:53