0000290: Loss of Density in Heavy Atom Distances of 2.855A

View Issue Details [ Jump to Notes ]

[ Issue History ] [ Print ]

Project

Category

View Status

Date Submitted

Last Update

0000290

Rosetta

[All Projects] Incorrect Results

public

2013-07-09 17:06

2013-07-15 10:26

Reporter

momeara

Assigned To

AndrewLeaverFay

Priority

normal

Severity

major

Reproducibility

always

Status

resolved

Resolution

fixed

Platform

All platforms

Any

OS Version

Any

Product Version

Trunk

Fixed in Version

Trunk

Summary

0000290: Loss of Density in Heavy Atom Distances of 2.855A

Description

There seems to be a loss of density at heavy atom distances of 2.855A (At least HBonding heavy atom Acceptor - Donor distances). It seems to be caused by one of these flags:
    -corrections::chemical:icoor_05_2009
    -analytic_etable_evaluation
    -use_bicubic_interpolation
    -dun10
    -set_atom_properties fa_standard:ONH2:LK_DGFREE:-5.85 fa_standard:NH2O:LK_DGFREE:-7.8 fa_standard:Narg:LK_DGFREE:-10.0 fa_standard:OH:LK_DGFREE:-6.70

It is apparent in bb->bb and sc->sc hbonds so my guess is that it is caused by a concave derivative discontinuity in the analytic_etable evaluation code.

Evidence: for plots, see attached tar.gz


For Talaris2013 (trunk) energy function, For Asp/Glu -> Lys hbonds:
     top8000 relaxed natives with no energy function flags
     O->N distance distribution with a tight kernel shows a striking loss of density at 2.855A
          Talaris2013_CXL_AMO_hbond_ADdist_130709.pdf
     O->N vs O->H scatter plot, shows the loss is correlated with the O->N distance
          hbond_geo_dim_pairs_scatter_ADdist_AHdist_chem_type_Talaris2013_130709.pdf

Score12 energy function, backbone->backbone hbonds:
      top8000 relaxed natives with flags:
           -restore_pre_talaris_2013_behavior
      O->N distance distribution with tight kernel zoomed in shows no loss of density at 2.855A
           hbond_ADdist_bb_bb_zoom_Score12_130709.pdf

Baseline energy function, backbone->backbone hbonds:
      top8000 relaxed natives with flags:
          -restore_pre_talaris_2013_behavior
          -corrections::chemical:icoor_05_2009
          -analytic_etable_evaluation
          -use_bicubic_interpolation
          -dun10
          -set_atom_properties fa_standard:ONH2:LK_DGFREE:-5.85 fa_standard:NH2O:LK_DGFREE:-7.8 fa_standard:Narg:LK_DGFREE:-10.0 fa_standard:OH:LK_DGFREE:-6.70
            <s weights=score12prime>
                <Reweight scoretype=fa_sol weight=0.75/>

                This was mistakenly set to 0.575 which is not .5 * 1.17
                <Reweight scoretype=hbond_sr_bb weight=0.585/>

                Enable the new disf
                <Reweight scoretype=dslf_ss_dst weight=0/>
                <Reweight scoretype=dslf_cs_ang weight=0/>
                <Reweight scoretype=dslf_ss_dih weight=0/>
                <Reweight scoretype=dslf_ca_dih weight=0/>
                <Reweight scoretype=dslf_fa13 weight=1.0/>
        </s>
      O->N distance distribution with tight kernel zoomed in shows loss of density at 2.855A
           hbond_ADdist_bb_bb_zoom_Baseline_130709.pdf

Steps To Reproduce

I see the problem in relax natives

Tags

No tags attached.

Application(s) Affected

rosetta_scripts FastRelax standard mover

Command Line Used

standard rosetta_scripts command line

Developer Options

Confirmed As Bug

Fixed in SVN Version

55496

Attached Files

long_dip.zip [^] (231,129 bytes) 2013-07-09 17:06

Relationships

Notes
(0000305) AndrewLeaverFay (Administrator) 2013-07-15 10:26	https://github.com/RosettaCommons/main/compare/5e4b5398ab...234478ca15 [^]

Issue History
Date Modified	Username	Field	Change
2013-07-09 17:06	momeara	New Issue
2013-07-09 17:06	momeara	Status	new => assigned
2013-07-09 17:06	momeara	Assigned To	=> AndrewLeaverFay
2013-07-09 17:06	momeara	File Added: long_dip.zip
2013-07-15 10:26	AndrewLeaverFay	Fixed in SVN Version	=> 55496
2013-07-15 10:26	AndrewLeaverFay	Note Added: 0000305
2013-07-15 10:26	AndrewLeaverFay	Status	assigned => resolved
2013-07-15 10:26	AndrewLeaverFay	Fixed in Version	=> Trunk
2013-07-15 10:26	AndrewLeaverFay	Resolution	open => fixed