You are here

very good I_sc values (if not too good)

3 posts / 0 new
Last post
very good I_sc values (if not too good)
#1

I use Rosetta 3.5 to perform SymDock using D8 symmetry and got most docking decoys of the following scores:

SCORE:     score     fa_atr     fa_rep     fa_sol    fa_elec    fa_pair    hbond_sr_bb    hbond_lr_bb    hbond_bb_sc    hbond_sc    dslf_ss_dst    dslf_cs_ang    dslf_ss_dih    dslf_ca_dih    atom_pair_constraint     fa_dun       I_sc    cen_rms        rms    st_rmsd     maxsub       time     description
SCORE:   781.142  -6075.987    209.017   2592.860    -53.066    -77.647       -235.804       -163.348        -93.127     -83.263       4189.354         46.876         99.084          2.994                 218.498    204.702 -273568608.403      0.000      0.000      0.000    234.000   1172.000 MNMR_input_0001

 

the I_sc values seems to be too good. It is also amazing that the total score is still positive with such good I_sc scores. I relaxed the unit before doing symmetric docking. No obvious clashes are observed while inspecting the docking results. What could be the reason for the I_sc values?

Category: 
Post Situation: 
Tue, 2015-11-03 03:30
attesor

Off the cuff, I'm guessing Rosetta thinks you have a disulfide between the two binding partners. Rosetta doesn't break the disulfide before calculating the "apo" reference score, so when you translate the two binding partners apart, you're left with a disulfide bond with a multi-hundred Angstrom long bond. This results in a very high positive (bad) energy for the apo state, so the moderately decent bound state looks *exceedingly* good in comparison.

Tue, 2015-11-10 11:14
rmoretti

yes, you are right that the dslf_ss_dst term is very positive.  There are cys residues in my protein but they do not form SS bonds. It seems in other results Rosetta correctly sees that and leaves the  dslf_ss_dst  term zero.  I specified -detect_disulf false but the dslf_ss_dst term is always positive in the score file. SS bonds were always detected. So I looked at source code:

core/pose/util.cc

        if ( option[ in::detect_disulf ].user() ?
                option[ in::detect_disulf ]() : // detect_disulf true
                pose.is_fullatom() // detect_disulf default but fa pose
            ) {
            pose.conformation().detect_disulfides();
        }
    }

core/pose/symmetry/util.cc

    if ( pose_out.is_fullatom() ) {
        pose_out.conformation().detect_disulfides();
    }

So it seems for full-atom models SS bonds are always detected.

But what I am docking is a hexamer and SS bonds reported by Rosetta are 80 Angstroms long !!! Though Rosetta says with mutual distance: 1.588 and 2.343:

core.conformation.symmetry.Conformation: Found disulfide between residues 3 1568
core.conformation.symmetry.Conformation: current variant for 3 CYS
core.conformation.symmetry.Conformation: current variant for 1568 CYS
core.conformation.symmetry.Conformation: current variant for 3 CYD
core.conformation.symmetry.Conformation: current variant for 1568 CYD
core.conformation.symmetry.Conformation: Found disulfide between residues 164 1407
core.conformation.symmetry.Conformation: current variant for 164 CYD
core.conformation.symmetry.Conformation: current variant for 1407 CYD
core.conformation.symmetry.Conformation: current variant for 164 CYD
core.conformation.symmetry.Conformation: current variant for 1407 CYD
core.conformation.Conformation: Connecting residues: 471 ( CYD ) and 2036 ( CYD ) at atoms  SG  and  SG 
core.conformation.Conformation:  with mututal distances: 2.343 and 1.58843
core.conformation.Conformation: Connecting residues: 632 ( CYD ) and 1875 ( CYD ) at atoms  SG  and  SG 
core.conformation.Conformation:  with mututal distances: 1.58843 and 2.343
core.conformation.Conformation: Connecting residues: 939 ( CYD ) and 2504 ( CYD ) at atoms  SG  and  SG 
core.conformation.Conformation:  with mututal distances: 2.343 and 1.58843
core.conformation.Conformation: Connecting residues: 1100 ( CYD ) and 2343 ( CYD ) at atoms  SG  and  SG 
core.conformation.Conformation:  with mututal distances: 1.58843 and 2.343

btw,  mututal should be a typo. 

Tue, 2015-11-17 07:36
attesor