The scripts and input files that accompany this demo can be found in the demos/public directory of the Rosetta weekly releases.

KEYWORDS: NUCLEIC_ACIDS EXPERIMENTAL_DATA RNA DENOVO STRUCTURE_PREDICTION

Written in March 2018 by Kalli Kappel (kappel at stanford dot edu). Updated July 2018.

This documentation has been verified to be compatible with Rosetta weekly releases: 2018.12, 2018.17, 2018.19, 2018.21, and 2018.26.

This demo shows how to use DRRAFTER to build a structure of an RNA-protein complex into a cryoEM density map and how to estimate the error in the resulting models.

Installing DRRAFTER:

  1. Download Rosetta here. You will need to get a license before downloading Rosetta (free for academic users). DRRAFTER is available in the Rosetta weekly releases starting with 2018.12. DRRAFTER is NOT available in Rosetta 3.9.
  2. If you're not using the precompiled binaries (these are available for Mac and Linux and you can access them by downloading source+binaries in Step 1), install Rosetta following the instructions available here.
  3. Make sure that you have python (v2.7) installed.
  4. Install Rosetta RNA tools. See instructions and documentation here.
  5. Check that the ROSETTA environmental variable is set (you should have set this up during RNA tools installation). Type echo $ROSETTA. This should return the path to your Rosetta directory. If it does not return anything, go back to step 4 and make sure that you follow the steps for RNA tools setup.
  6. Set up the executables for DRRAFTER (don't worry about "File exists" warnings from these commands). Note: the following commands are for bash, if you are using a different shell, you will need to modify these commands accordingly. Check which shell you're using by typing echo $0. If you're using bash, type: 
    ln -s $(ls $ROSETTA/main/source/bin/rna_denovo* | head -1 ) $ROSETTA/main/source/bin/rna_denovo
    Then type: 
    ln -s $(ls $ROSETTA/main/source/bin/drrafter_error_estimation* | head -1 ) $ROSETTA/main/source/bin/drrafter_error_estimation
    Finally, type: 
    ln -s $(ls $ROSETTA/main/source/bin/extract_pdbs* | head -1 ) $ROSETTA/main/source/bin/extract_pdbs
  7. Add the path to the DRRAFTER script to your $PATH (alternatively, you can type the full path to the DRRAFTER.py script each time that you use it). It is found in main/source/src/apps/public/DRRAFTER/ in your Rosetta directory. An example for bash: 
    export PATH=$PATH:$ROSETTA/main/source/src/apps/public/DRRAFTER/

Brief explanation of input files:

All of the necessary files for this demo are available in $ROSETTA/demos/public/drrafter/, where $ROSETTA is the path to your Rosetta installation.

fasta.txt: The FASTA file listing the full sequence of the complex being modeled. It should contain at least one line that starts with '>' and lists chains and residue numbers for the sequence, e.g. A:136-258 E:1-23. Here we are modeling chain A residues 136-258 and chain E residues 1-23. The subsequent lines should list the full sequence of the complex. Protein residues are specified by uppercase one-letter codes. RNA residues are specified with lowercase one-letter codes ('a', 'u', 'g', and 'c'). Protein residues should be listed before RNA residues.

secstruct.txt: A file containing the secondary structure of the complex in dot-bracket notation. Secondary structure for the protein should be specified by dots. The secondary structure should be the same length as the sequence found in the fasta file. For RNA residues, this secondary structure will be enforced during the DRRAFTER run. RNA secondary structures can be predicted computationally with packages such as ViennaRNA. If the secondary structure is not known, it may be necessary to test several different secondary structures in separate DRRAFTER jobs (or ideally the secondary structure would be determined through biochemical experiments).

1wsu_simulated_7A.mrc: The density map file in mrc format (ccp4 is also acceptable). For this demo, this map has been simulated from PDB ID 1WSU at 7Å.

RNA_helix.pdb: This is an ideal RNA helix corresponding to residues E:1-4 and E:20-23. Ideal RNA helices can be generated using rna_helix.py (documentation on the RNA tools page). This PDB was generated with the following command: 

rna_helix.py -seq ggcg cgcc -o RNA_helix.pdb -resnum E:1-4 E:20-23 -extension static.linuxgccrelease
Note that you may need to change the extension in the above command (see instructions in the RNA tools documentation). PDBs used in DRRAFTER must have chain IDs (i.e. you cannot provide -resnum 1-4 20-23 to rna_helix.py to generate an RNA helix for DRRAFTER).

protein_fit_into_density.pdb: The protein structure that has been fit into the density map. For this demo, this is a crystal structure of the unbound protein (PDB ID 1LVA). The sequence of this protein structure should exactly match the sequence in the fasta file.

protein_and_RNA_helix_fit_into_density.pdb: Coordinates of both the protein structure and the RNA helix (from RNA_helix.pdb) fit into the density map. This will be the starting conformation for the DRRAFTER run. (The relative positions of the protein and RNA helix will be allowed to change though.) The order of the residues in this PDB file should match the order of the residues in the fasta file.

Running DRRAFTER:

1. Go to the DRRAFTER demo directory. Type: 

cd $ROSETTA/demos/public/drrafter/

2. Use DRRAFTER.py to set up the run. Type:

DRRAFTER.py -fasta fasta.txt -secstruct secstruct.txt -start_struct protein_and_RNA_helix_fit_into_density.pdb -map_file 1wsu_simulated_7A.mrc -map_reso 7.0 -residues_to_model E:1-23 -include_as_rigid_body_structures protein_fit_into_density.pdb RNA_helix.pdb -absolute_coordinates_rigid_body_structure protein_fit_into_density.pdb -job_name demo_run -dock_into_density -demo_settings -rosetta_directory $ROSETTA/main/source/bin/

Note that -rosetta_directory needs to provide the path to your Rosetta executables (if you installed DRRAFTER following the instructions above, then $ROSETTA/main/source/bin/ should be fine). If the path to the Rosetta executables is already in your system PATH, then you can omit the -rosetta_directory flag.
Note also the -demo_settings flag: this flag is designed to make the DRRAFTER run finish quickly, and should not be used for normal runs. Specifically, -demo_settings is equivalent to the following options: -cycles 500 -extra_flags 'rnp_high_res_cycles 0' 'minimize_rounds 1' 'no_filters' 'nstruct 10'.
For normal runs: the number of structures built per DRRAFTER job can be set with e.g. -extra_flags 'nstruct 2000'.

This will create the following files:

fasta_demo_run.txt: The FASTA file for the region that will be included in the DRRAFTER run.

secstruct_demo_run.txt: The file specifying the secondary structure for the region that will be included in the DRRAFTER run.

coord_csts_demo_run.txt: A file describing coordinate restraints that will be applied during the DRRAFTER run. By default, residues in the starting structure will be restrained to be within 10 Å of their initial coordinates. This can be turned off with -no_csts and the distance at which the restraints will be activated (default 10 Å) can be controlled with -cst_dist.

flags_demo_run: A file listing all of the Rosetta options for the run.

init_struct_demo_run.pdb: The starting structure for the DRRAFTER run. In this case, this structure is identical to protein_and_RNA_helix_fit_into_density.pdb.

DRRAFTER_command: This file contains the command to run the DRRAFTER job.

3. Run the DRRAFTER_command. This can be done either by typing: 

source ./DRRAFTER_command

OR by copying the line in the DRRAFTER_command file to the command line:

/your/path/to/rosetta/executables/rna_denovo @flags_demo_run

This will take several minutes to run and it should create a file called demo_run.out, which contains all of the structures from the run.

4. Extract PDB files from the compressed output file created in the previous step. Type: 

extract_lowscore_decoys.py demo_run.out 10

This extracts the 10 best scoring structures from the run and will create 10 PDB files named demo_run.out.1.pdb, demo_run.out.2.pdb, etc. Note that in this case we only generated 10 structures total, but for a real run it is recommended that you build at least 2000-3000 structures. This can be done by setting the number of structures built per DRRAFTER run with -extra_flags 'nstruct 2000' (if this option isn't provided and the -demo_settings option isn't supplied, then default=500).

5. Look at the structures! Open them in pymol or Chimera. You will see that all of the missing RNA residues have been built. The protein and RNA helix have also moved slightly from their initial positions.

6. Estimate the error in the DRRAFTER models. Again, use the DRRAFTER.py script to do this. Type: 

DRRAFTER.py -final_structures demo_run.out.1.pdb demo_run.out.2.pdb demo_run.out.3.pdb demo_run.out.4.pdb demo_run.out.5.pdb demo_run.out.6.pdb demo_run.out.7.pdb demo_run.out.8.pdb demo_run.out.9.pdb demo_run.out.10.pdb -estimate_error -rosetta_directory $ROSETTA/main/source/bin/

This will print out information about the error estimation to your screen. For example (the actual numbers may vary!): 

apps.public.DRRAFTER.drrafter_error_estimation: #############################################
apps.public.DRRAFTER.drrafter_error_estimation: Mean pairwise RMSD (convergence): 5.41042
apps.public.DRRAFTER.drrafter_error_estimation: Estimated minimum RMSD: 3.65777
apps.public.DRRAFTER.drrafter_error_estimation: Estimated mean RMSD: 5.39297
apps.public.DRRAFTER.drrafter_error_estimation: Estimated RMSD of median structure: 4.73551
apps.public.DRRAFTER.drrafter_error_estimation: Median structure: demo_run.out.1.pdb
apps.public.DRRAFTER.drrafter_error_estimation: #############################################

All numbers have units of Å. The mean pairwise RMSD describes the "convergence" of the run, i.e. how similar the final structures are to each other. The estimated RMSD (root mean square deviation) values to the "true" coordinates are based on this convergence value. The estimated minimum RMSD predicts the best accuracy of the final structures. The estimated mean RMSD predicts the average RMSD accuracy of the final structures. The median structure is determined to be the final structure with the lowest average pairwise RMSD to the other final structures. The accuracy estimate of this model is also printed to the screen.

For reference, example output is provided in the example_output/ directory.

Additional information

See the DRRAFTER documentation here.