3 Steps to a CS-Rosetta structure

Running CS-Rosetta is made simple by using the provided automatic setup tools from the CS-Rosetta Toolbox. Standard settings are provided and give good results in a broad range of cases. Tools are provided to prepare (convert, trim) the input files: chemical shifts, distance restraints, NOE peak lists, PCS data and RDCs.

Step 1: Fragment Picking

First you need to use the backbone chemical shifts to pick CS-Rosetta fragments. This is carried out using the pick_fragments application. If your sequence contains flexible tails (e.g., his-tags) it can be advantageous to trim these before you commence with fragment picking. Details are explained in Tutorial: Fragment Picking with Chemical Shift Data.

Step 2: Run Simulation

Use the setup_target and setup_run applications of the Toolbox to setup a complete Rosetta run directory. The -method option allows you to choose between different algorithms such as abrelax (standard rosetta), RASREC (recommended) and AutoNOE (experimental/unpublished). The protocols are described in detail in the Manual Sections: CS-Rosetta, RASREC and AutoNOE. A step-by-step how to for running the basic algorithms can be found in Tutorial: Standard CS-Rosetta, Tutorial: RASREC CS-Rosetta, and Tutorial: AutoNOE CS-Rosetta. For RASREC and abrelax the choice of the algorithm does not change how additional restraint data is handled. For automatic NOESY assignment you have to run AutoNOE CS-Rosetta which builds upon the iterative RASREC protocol. How to add restraint data is described in Tutorial: Calculations with Restraints

Step 3: Post Processing

CS-Rosetta will produce a lot of structures that are usually saved in silent file format. Use the provided tools (extract_decoys, extract_scores and extract_pdb) to extract low-energy structural ensembles from the large collection of structures returned by ROSETTA. If you want to add additional scores to the final structures this is best done by applying the score_jd2 application directly to the raw silent-IO file returned by ROSETTA. In addition to ROSETTA energies score_jd2 can compute SPARTA+ chemical shift scores, restraint scores (distance restraints, RDC and PCS restraints) and RMSDs. See details in Tutorial: (Re-)Scoring of Structures.


Hi Oliver,

The page "Tutorial: (Re-)Scoring of Structures" is linking to "3 Steps to a CS-Rosetta structure". Could you please correct it?

What is the "chem_shifts" term in the output scores? Is it the same as 0.25*Chi^2 in the original CS-rosetta paper?


Yisong Tao

Hi Yisong,

the link is corrected, but the tutorial hasn't been written yet.

Yes, the chem_shifts term corresponds to the same score as in the original paper, but using SPARTA+ for the calculation of shifts from structure.



Is there a way to remove homologous sequences during fragment picking?

Thank you,

Hi Brinda,

the flag -nohom should be enough do remove these sequences, but there is currently a bug in the fragment picker, which will be removed with the next release.

For now some useful commands to check and remove manually:
The file t000_.fasta.homologous is already generated by the fragment picker and should be found in the directory with your fragments:

write the following lines to a script: 'homs_grep.sh'
cat $frags | gunzip | egrep "$( cat $homs | awk '{print $NF}' | sed 's/\(....\)\(.\)/\1 \2/ ' | awk '{printf("%s\\ %s|",$1,$2)}' | sed 's/|$//' )" | cut -f 2 -d " " | uniq

# then this command shows you the homologous that still appear in your fragments.
$ hom_grep.sh t000_.fasta.frags9.gz t000_.fasta.homologous

write the following lines to a script 'remove_homs.sh'
cat $frags | gunzip | egrep -v "$( cat $homs | awk '{print $NF}' | sed 's/\(....\)\(.\)/\1 \2/ ' | awk '{printf("%s\\ %s|",$1,$2)}' | sed 's/|$//' )" | gzip > $(basename $frags .gz)_nohom.gz

# then this command should generate a new fragment file without any homs
$ remove_homs.sh 000_.fasta.frags9.gz t000_.fasta.homologous


The (Re-)Scoring of Structures link is now updated.