Rosetta documentation page

This page used to provide information and command lines for Rosetta. Most of the information was included into the official Rosetta documentation website, which is a vast and comprehensive source of information about the software. Developers' version of this documentation may be accessed by following this link (Rosetta developer’s github account required)

Here we provide additional infomation and input files regarding the protocols frequently used in our lab.

Before you start

  • Any Rosetta executable program has its name and an extension. The extension depends on the type of you build (either release or debug), your architecture (e.g. macos, linux, etc) and compiler (e.g. gcc or icc). The extension may look like one of the following: macosclangrelease, linuxgccrelease or linuxgccdebug

Installing Rosetta

Follow this link

Frequently used protocols


The purpose of idealization is to alter bond lengths and planar angles so they agree with the Rosetta ideal values. Unless you do a high resolution comparative modeling or you want to redesing a protein with known crystal structure, you should idealize your input PDB files.



Ab-initio protocol predicts protein structures in the absence of any homologous protein of known 3D coordinates.

Workflow: 1. Save the sequence of your protein target in fasta format. 2. Prepare Rosetta-fragments using fragment_picker. 3. Ab-initio modeling using minirosetta. 4. Select the best N models by score, where N is typically 1000 - 20 000 5. Extract these models by extract_pdbs.
6. Run clustering and further analysis.

Fragment Picker

Fragment picking has been covered in this document.


This program (minirosetta) runs runs different modes or protocols (classic abinitio and relax) depending on command line arguments for de novo protein fold prediction.


    ~/PATH/minirosetta.VERSION @NAME-OF-FILE-WITH-FLAGS.flags_abinitio


    ~/Rosetta_src/main/source/bin/minirosetta.linuxgccrelease @QUERY.flags_abinitio

flag file
This file contains all necessary flags with parameters that are needed to start de novo modeling of protein structure.
e.g. query_2gb1.flags_abinitio:

-sep_switch_accelerate 0.8
-no_minimize false
-abinitio::rsd_wt_helix 0.5
-abinitio::rsd_wt_loop 0.5
-abinitio::rg_reweight 0.5
-abinitio::use_filters false
-checkpoint_interval            #[-checkpoint] - turn checkpointing on with given time interval in seconds
-abinitio::relax true           #[-abinitio::relax <bool>] - do relax after abinitio
-nstruct 5                      #[-nstruct <int>] - number of decoys (default 1)
-increase_cycles 10             #[-increase_cycles <float>] - increase no. of cycles at each stage by this factor
-frag3 QUERY.200.3mers
-frag9 QUERY.200.9mers
-in::file::fasta QUERY.fasta
-in::file::silent QUERY.out     #[-out::file::silent <silent file name>] - name of silent output file
-out::silent_gz                 #[-out::silent_gz] - gzip silent output file
-psipred_ss2 QUERY.ss2
-in:path:database /PATH/Rosetta/database


  • QUERY.fasta - the same as above
  • QUERY.ss2 - the same as above
  • QUERY.200.3mers - the same as above
  • QUERY-frags.3.scf - the same as above
  • QUERY.flags_abinitio - file with flags for minirosetta


  • QUERY.out - output file in silent format of data; contains all decoys from a job
  • score.fsc - output file contains SCORE for all decoys from silent-file


This application can be used to extract PDBs (retrieve pdb-files) from silent-files.


    ~/PATH/extract_pdbs.linuxgccrelease @NAME-OF-FILE-WITH-FLAGS.flags_extract


    ~/Rosetta_src/main/source/bin/extract_pdbs.linuxgccrelease @query_2gb1.flags_extract

This file contains all necessary flags with parameters that are needed to retrieve pdb-files from silent-file.
e.g. query_2gb1.flags_extract:

-in::file::silent QUERY.out             #Path and names, specifies list of silent-files to be read in.
-in:file:tags YOUR_TAG                  #List of tags to be extracted from silent-files.
-in::file::silent_struct_type protein   #Strucutre type of input, choices: rna, binary, protein, pdb.
-in::file::residue_type_set             #Used in creating Pose from silent-files: centroid, fa_standard.
-out::prefix YOUR_PREFIX                #Prefix that is appended onto extracted PDB files.
-in::file::rescore true/false           #Governs whether Pose from silent-file is rescored or not.
-score::weights                         #Weights file for ScoreFunction, defaults to standard.
-score::patch                           #Patch file for ScoreFunction.


  • QUERY.out - the same as above
  • query_2gb1.flags_extract - file with flags for extract_pdbs


  • model_number.pdb - lots of pdb-files with extract models, each in a separate file