Validation of non-protein/non-NA atomic structures

[aozalevsky]

Entry 215 presented a challenge for the validation pipeline. Because it's a full-atom structure, we used standard Molprobity analysis. However, because there were no proteins or nucleic acids the report came out essentially blank (no Ramachandran or side-chains outliers).

============================ Molprobity validation ============================   
                                                                               
                                                                                                                                                               
  Geometry Restraints Library: GeoStd + Monomer Library                        
  Deviations from Ideal Values - rmsd, rmsZ for bonds and angles.            
    Bond      :  0.224   6.733  80239  Z=16.757                                                                                                                
    Angle     :  6.278  97.518 108743  Z= 2.093                                                                                                                
    Chirality :  0.122   0.709  28044                                                                                                                          
    Planarity :  0.104   0.326    460
    Dihedral  : 13.506 149.932 132182                                                                                                                          
    Min Nonbonded Distance : 1.194                                                                                                                             
                                                                                                                                                                                                                                                                                                                              
  Molprobity Statistics.                                                                                                                                       
    All-atom Clashscore : None                                                                                                                                 
    Ramachandran Plot:                                                                                                                                         
      Outliers :  0.00 %                                                                                                                                       
      Allowed  :  0.00 %                                                                                                                                       
      Favored  :  0.00 %                                                                                                                                       
    Rotamer:                                                                                                                                                                                                                                                                                                                  
      Outliers :  0.00 %                                                                                                                                                                                                                                                                                                      
      Allowed  :  0.00 %                                                       
      Favored  :  0.00 %                                                                                                                                                                                                                                                                                                      
    Cbeta Deviations :  0.00 %                                                                                                                                 
    Peptide Plane:                                                                                                                                                                                                                                                                                                            
      Cis-proline     : 0.00 %                                                                                                                                                                                                                                                                                                
      Cis-general     : 0.00 %                                                                                                                                 
      Twisted Proline : 0.00 %                                                                                                                                                                                                                                                                                                
      Twisted General : 0.00 %                                                                                                                                                                                                                                                                                                
                                                                                                                                                               
  Rama-Z (Ramachandran plot Z-score):                                                                                                                                                                                                                                                                                         
  Interpretation: bad |Rama-Z| > 3; suspicious 2 < |Rama-Z| < 3; good |Rama-Z| < 2.                                                                                                                                                                                                                                           
  Scores for whole/helix/sheet/loop are scaled independently;                                                                                                  
  therefore, the values are not related in a simple manner.                                                                                                                                                                                                                                                                   
    whole:  None (None), residues: 0                                                                                                                                                                                                                                                                                          
    helix:  None (None), residues: 0                                                                                                                                                                                                                                                                                          
    sheet:  None (None), residues: 0                                                                                                                                                                                                                                                                                          
    loop :  None (None), residues: 0                                                                                                                                                                                                                                                                                          
                                                                                                                                                               
  Max deviation from planes:
     Type  MaxDev  MeanDev LineInFile

=================================== Summary =================================== 

  Ramachandran outliers =   0.00 %
                favored =   0.00 %
  Rotamer outliers      =   0.00 %
  C-beta deviations     =     0
  RMS(bonds)            =   0.2235
  RMS(angles)           =   6.28
  MolProbity score      =  -1.00

Moreover, the clashscore calculation also failed:

Bad Clashes >= 0.4 Angstrom:                                                                                                                                                                                                                                                                                                  
Sorry: PROBE output is empty. Model is not compatible with PROBE.  

One way to approach such entries in the future might be to treat them as CG and do an excluded volume calculation. To do that we have to figure out two issues:

1. choose an appropriate atomic radii database, because atom_site table doesn't provide radii

2. tune the performance of the excluded volume calculation algorithm, because it wasn't designed to handle 250k+ particles systems.