You are here

RosettaAntibodyDesign: How can I run the protocol without allowing design?

6 posts / 0 new
Last post
RosettaAntibodyDesign: How can I run the protocol without allowing design?

Hi there,

I am trying to use RosettaAntibodyDesigner without allowing any design elements of the protocol to occur. I understand 100% that this would remove the intended purpose and use cases of RAbD, however, I am trying to piecemeal a side-by-side comparison of a designed mAb (which works reat through this protocol) and a non-designed mAb. 

However, I would like to ensure that both my simulations are executred identially otherwise. (ie., loop flexibility in docking, etc).

Is there any way to runn RAbD while FIXing all the CDRs?

My command is:

$ROSETTA_BIN/antibody_designer.macosclangrelease \
  -database $ROSETTA_DATABASE \
  -do_dock   \
  -use_epitope_constraints \
  -in:file:s complex_start.pdb \
  -cdr_instructions instr.txt 

My instr.txt file is:

# No design
ALL GraftDesign FIX
ALL SeqDesign FIX
ALL MinProtocol MINTYPE relax


And the error I am seeing is:

ERROR: No CDRs to design!  Please set CDRs via instructions file, cmd-line, or RosettaScripts!


Is there any way to have RAbD run without implementing the grafting and sequence design steps?


Post Situation: 
Tue, 2020-05-26 08:12



There is currently no way to do this WITHIN the protocol itself for a given antibody.    Honestly, it's a monte-carlo algorithm, so it would not be possible to directly compare one to the other.  There are work-arounds though. 

1) GraftDesign - if your antibody came from the PDB, you can set the CDRSet in the instructions file to only graft from that PDBId.  Therefore the sequence will stay the same.  There is currently no way to give arbitrary PDB files and insert them into the database. (you could of-course hack this using sqlite3 module of python to add your antibody into the db.)

2) SeqDesign - you can set the instruction file to have "NONE" for design of each CDR.  You still need to give the CDRs you will design for the control as you do for the design. 

3) RosettaScripts - this is probably your best bet.  You can match roughly what is going on with packing and backbone minimization for a better comparison through the RosettaScript components.   But, this is not trivial and basically involves re-creating the protocol step-by step in RS.  You're probably better off relaxing your starting antibody with CDR dihedral constraints, and coordinate constraints on the framework, while doing the same for the design.  Then, you'll have a fairly decent comparison between the two.

In general - for your starting antibody, you want to have this antibody minimized with constraints.  Typically, I use the pareto-optimal method by Lucas Nivan and Rocco Moretti. However, if you did this to your starting antibody with CDR cluster dihedral constraints instead (there is a ConstraintGenerator for this), then you will have a better place to start from for comparisons. 



Tue, 2020-05-26 08:37

This is super helpful, thank you.

Tue, 2020-05-26 08:48

I'm finally getting around to trying this out and am seeing some issues. I am trying option (2) since I am not using a starting ab from the pdb (excluding option 1) and the short term earning curve for option 3 is too large (for now). In attempting (2):

my instr.txt file now looks like:

# No design; just flexible motion of the loops
# The starting ab is a camelid, so no light chains.

ALL GraftDesign FIX

ALL MinProtocol MINTYPE relax

and I see the error:

File: src/protocols/antibody/design/
[ ERROR ] UtilityExitException
ERROR: Unrecognized option: NONE

Any ideas how I specifiy the "NONE" that you recommended?


Tue, 2020-06-02 12:52

Sorry about that.  Looks like  I don't talk about it much in the documentation. It says it works for fallback strategy, but in fact, it will completely disable design of a particular CDR if you give it for both main and fallback strategy. 


You'll want to set the seqdesign strategy to 'NO_DESIGN' or 'OFF'. 

So in your instruction file, it will look this:

L1 SeqDesign Strategy OFF  
L1 SeqDesign Fallback_Strategy OFF
Tue, 2020-06-02 14:00

Ah great. Thank you.

I found that adding the disallow_aa term with all the AA's to the command line worked.

-disallow_aa  A R N D C Q E G H I L K M F P S T W Y V

I'll give your suggestions a shot though too, since I think the above is pretty hacky.

Thanks again!

Wed, 2020-06-03 13:20