Loop Modeling Application

This release of Rosetta3 includes updates to the Rosetta 2.x cyclic coordinate descent loop modeling. You can find the code for the loop modeling protocols in src/apps/public/loop_modeling/. The executable is located at bin/loopmodel.*. A sample package is in ReleaseSamples/Loop/.

Input Files

1. Start pdb
2. Loops file

   Here is the loop file format. The columns are whitespace delimited. Only 1-3 are necessary.

   column1  "LOOP":     Literally the string LOOP, identifying this line as a loop
                        In the future loop specification files may take other data.
   column2  "integer":  Loop start residue number
   column3  "integer":  Loop end residue number
   column4  "integer":  Cut point residue number, >=startRes, <=endRes. Default or 0: let LoopRebuild choose cutpoint randomly.
   column5  "float":    Skip rate. default - never skip
   column6  "boolean":  Extend loop. Default false
   

   For Example: A legal line in a loops file (the line by itself is a legal file):

   LOOP 23 30 26
   Loop from residues 23-30 with the cutpoint at 26.
   

3. Fragments file(s), as needed. Fragments are used to speed sampling and capture local elements of secondary structure.

Command Lines and Options

1. Sample Command:

   ./loopmodel.linuxgccrelease @flags 

2. Options used in Loop Modeling
   • General inputs

     -nstruct                  number of outputs. [Integer]
     -database                 Path to rosetta3 databases. [PathVector]
     -in:file:fullatom         Enable full-atom input from PDB. [Boolean]
     -loops::input_pdb         Path/name of input pdb file.  Note that this is NOT -s/-l like normal. [File]
     -loops:loop_file          Path/name of loop definition file. default "loop_file". [File]
     -loops::frag_sizes        Length of fragments to be used in loop modeling (in other words, number of amino
                               acids/residues for each segment. For each type of fragment). defaults to 9 3 1 [IntegerVector]
     -loops::frag_files        Path/name of fragment libraries files, defaults to frag9 frag3 frag1. [FileVector]
     

   • Secondary structure options

     Rosetta can figure out secondary structure from phi/psi, but if you know it already (and/or trust your assignment more than Rosetta's) you can use these flags:

     -in::file::psipred_ss2    For psipred_ss2 (pass path to file). [File]
     -in::file::dssp           For dssp (pass path to dssp file). [File]
     

   • Loop modeling control

     A series of string options control what sorts of loop modeling you get. The executable contains many different loop modeling modes, you use string selections to tell it which paths to take. These options are listed in the order that they take effect (the order that the steps happen in the program).

     -loops::build_initial     Pre-build the loops to ensure missing atoms are accounted for, and no chainbreaks exist,
                               before modeling. [Boolean]
     -loops::remodel           Legal values: 'no','perturb_ccd','quick_ccd','quick_ccd_moves','old_loop_relax'.
                               default='quick_ccd'.[String]
                               This option controls the centroid-stage loop remodeling (where large perturbations are attempted).
     -loops::intermedrelax     Currently not used; eventually may provide for a full-pose relax between centroid and fullatom
                               modeling. default = 'no' [String]
     -loops::refine            Legal values: 'no','refine_ccd'. This option controls the fullatom refinement stage
                               of loop modeling. default = 'no' [String]
     
     -loops::idealize_before_relax
                               Idealize the structure after refinement, sometimes useful if you want a final relax step.[boolean]
     -loops::relax             Legal values: 'no','fastrelax','shortrelax','fullrelax'    [String]
                               Controls whether a full-structure relax occurs after loop modeling.  Defaults to "no".
     

Result Interpretation

In general, the result structures' energies are interpreted as normal. Because loop modeling extensively remodels the backbone of the loops, you will want to pay special attention to rama (Ramachandran energy) and omega (related to the likelihood of a given omega [peptide bond] dihedral; statistics from the PDB). It may be worth examining per-residue scores within the loops as well, because single residues' bad ramas can kill the plausibility of a loop structure. CCD-style loop modeling also has a chainbreak term (usually called break or cbreak or chain) which measures how well the artificial chainbreak introduced for CCD closure was re-annealed.

The specifically important terms:

rama             Ramachandran energy (phi and psi dihedrals in the backbone)
omega            omega dihedral in the backbone
chainbreak       measures whether the loop was successfully closed.

---