Ingemar André (
Sebastian Rämisch (

Corresponding PI: Ingemar André

Last edited 10/26/15.


  • Application source code: rosetta/main/source/src/apps/public/


This simple fragment set generator was first used in:

Sebastian Rämisch, Robert Lizatovic and Ingemar André,
Exploring alternate states and oligomerization preferences of coiled-coils by de novo structure modeling,
Proteins (2015);jsessionid=3B2AAEB6FBFDD42007B1A8DF5EABEA41.f04t03


Generate fragment sets for Rosetta protein structure modeling using a predefined set of protein structures as fragment source. The fragments are randomly picked from a set of structures, irrespective of sequence and secondary structure. The original intention was to predict coiled-coil structures using exclusively fragments taken from coiled-coil structures.


First, all possible fragments from a given set of PDB files are generated. Then, for each position of the target sequence, a set of fragments (default=200) is randomly picked. Secondary structure and sequence are not considered at any point.


  • It does not use the scoring scheme used by the standard fragment picker
  • This will not work with JD2

How to run

  1. Download all pdb files you intend to use as sources to some directory.
  2. Prepare a list file that contains path/name.pdb for each source structure (see below)
  3. Run struct_set_fragment_picker


The only input file is a list of the pdb names to be used. E.g.:



Option Explanation
-in::file::l list of structures
-frags:n_frags number of fragments per position (optional; default = 200)
-struc_set_fragment_picker::frag_length Fragment length. Usually 3 or 9.
-struc_set_fragment_picker::sequence_length Length of target sequence.
-struc_set_fragment_picker::frag_name Name of the output fragment file. Usually 5 characters.


The output is a standard Rosetta fragment file to be used for structure prediction.

New things since last release

This is the first public release

See Also