You are here

H3 modelling scoring function

3 posts / 0 new
Last post
H3 modelling scoring function

Dear all,

Let's suppose we have the following situation:

I have an antibody structure generated by the 'antibody.linuxgccrelease' (i.e. no H3 optimization yet, just grafting)

Is it possible to score it in a similar fashion as the 'antibody_H3.linuxgccrelease' application does? I mean to score it without, H3 optimization whatsoever?

I saw that in the scoring file of a H3 optimization run we can find different terms than in a normal Talaris2014 relaxing/scoring:

CDR_SASA CDR_SASA_HP CDR_charge     H1_RMS     H2_RMS     H3_RMS     L1_RMS     L2_RMS     L3_RMS VL_VH_distance .... etc.

Any help will be greatly appreciated.

Thank you,


Post Situation: 
Wed, 2017-12-27 23:10

Hey Andrei,

You can get approximate RMSDs by running, "antibody_H3.linuxgccrelease -antibody:model_h3 false -antibody:snugfit false -antibody:refine_h3 false -s model.pdb -native crystal.pdb". I say approximately becuase there are a few steps which cannot be skip in the antibody_H3 executable by setting options alone, so there will be a little motion (~0.1 A RMSD). Alterntively, I can provide a Python script that use PyRosetta to calculate loop RMSDs.

To get metrics regarding light–heavy chain orientation, please use the packing_angle.linuxgccrelease application. I believe all it requires is an input structure with "-s input.pdb".



Wed, 2018-01-03 08:16

Dear Jeliazko,

Many thanks for your reply! It works just fine. We already have a script in VMD to calculate the H3 RMSD and local fitting.

Do you think that it make any sense to calculate the sum of the scores for the H3 residues and use it instead of total score?


All the best,


Mon, 2018-01-08 23:49