Rosetta 3.15 - User Guide

Cluster: XLence (UNIMI Dipartimento di Scienze Farmacologiche e Biomolecolari) Date: October 12, 2025 Installation: /sw/rosetta/3.15_ARCH/ (ARCH = avx2 or avx512)

📦 Version Information

Rosetta: 3.15 (stable release) PyRosetta: 3.15 (Python bindings included) Python: 3.12 (miniforge3/20250911) Compiler: GCC 13.2 with AVX/AVX512 optimizations

Architecture Detection

The module automatically detects your CPU and loads the appropriate version: - AVX2 build: Login node - Intel i9-12900 - AVX-512 build: Compute nodes (node1-5, ngs) - Intel Xeon W-2155

🚀 Loading the Module

# For C++ applications only
module load rosetta/3.15

# For PyRosetta (Python bindings)
module load miniforge3
module load rosetta/3.15

# Verify loaded version
echo $ROSETTA_ROOT
echo $ROSETTA_ARCH      # Shows: avx2 or avx512
echo $ROSETTA_DATABASE

The module will automatically display which architecture was detected.

💻 Main Executables

Core Applications

Executable	Description
`rosetta_scripts`	General-purpose protocol executor (XML-based)
`score`	Energy scoring of structures
`relax`	Structure relaxation and refinement
`minimize`	Energy minimization

Protein Design

Executable	Description
`fixbb`	Fixed backbone design
`fast_design`	Sequence design with backbone flexibility
`remodel`	Loop modeling and design
`simple_cycpep_predict`	Cyclic peptide structure prediction

Docking & Interactions

Executable	Description
`docking_protocol`	Protein-protein docking
`FlexPepDock`	Peptide-protein docking
`antibody_H3`	Antibody CDR-H3 loop modeling

Structure Prediction

Executable	Description
`AbinitioRelax`	Ab initio structure prediction
`hybridize`	Template-based modeling
`loopmodel`	Loop modeling and refinement

Analysis Tools

Executable	Description
`extract_pdbs`	Extract structures from silent files
`score_jd2`	Batch scoring of structures
`InterfaceAnalyzer`	Protein interface analysis

🐍 PyRosetta (Python Interface)

PyRosetta provides a Python interface to all Rosetta functionality.

Loading PyRosetta

#!/usr/bin/env python3
import pyrosetta
pyrosetta.init()

# Load a pose
pose = pyrosetta.pose_from_pdb("input.pdb")

# Score the structure
scorefxn = pyrosetta.get_score_function()
score = scorefxn(pose)
print(f"Score: {score}")

Python Tools

Module	Description
`pyrosetta.rosetta`	Access to all C++ Rosetta objects
`pyrosetta.toolbox`	High-level Python utilities
`pyrosetta.teaching`	Educational tools and examples

📖 Usage Examples

1. Structure Relaxation (CPU)

#!/bin/bash
#SBATCH --job-name=rosetta_relax
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=8
#SBATCH --time=24:00:00

module purge
module load rosetta/3.15

# Relax structure
relax.linuxgccrelease \
  -s input.pdb \
  -relax:constrain_relax_to_start_coords \
  -nstruct 10 \
  -out:suffix _relaxed

2. Protein Design with RosettaScripts

#!/bin/bash
#SBATCH --job-name=rosetta_design
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=16
#SBATCH --time=48:00:00

module purge
module load rosetta/3.15

# Run design protocol
rosetta_scripts.linuxgccrelease \
  -parser:protocol design_protocol.xml \
  -s input.pdb \
  -nstruct 100 \
  -out:prefix design_ \
  -out:path:all output/

Example XML protocol (design_protocol.xml):

<ROSETTASCRIPTS>
  <SCOREFXNS>
    <ScoreFunction name="ref15" weights="ref2015"/>
  </SCOREFXNS>
  <TASKOPERATIONS>
    <InitializeFromCommandline name="init"/>
    <RestrictToRepacking name="rtr"/>
  </TASKOPERATIONS>
  <MOVERS>
    <PackRotamersMover name="design" scorefxn="ref15" task_operations="init"/>
    <MinMover name="min" scorefxn="ref15" chi="true" bb="false"/>
  </MOVERS>
  <PROTOCOLS>
    <Add mover="design"/>
    <Add mover="min"/>
  </PROTOCOLS>
</ROSETTASCRIPTS>

3. PyRosetta Script (Python)

#!/bin/bash
#SBATCH --job-name=pyrosetta_job
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=4
#SBATCH --time=12:00:00

module purge
module load miniforge3
module load rosetta/3.15

python3 my_design_script.py

Example script (my_design_script.py):

#!/usr/bin/env python3
import pyrosetta
from pyrosetta import rosetta

# Initialize PyRosetta
pyrosetta.init("-mute all")

# Load structure
pose = pyrosetta.pose_from_pdb("input.pdb")

# Setup score function
scorefxn = pyrosetta.get_score_function()

# Design protocol
task_factory = rosetta.core.pack.task.TaskFactory()
task_factory.push_back(rosetta.core.pack.task.operation.InitializeFromCommandline())

pack_mover = rosetta.protocols.minimization_packing.PackRotamersMover()
pack_mover.task_factory(task_factory)
pack_mover.score_function(scorefxn)

# Run design
for i in range(10):
    test_pose = pose.clone()
    pack_mover.apply(test_pose)
    score = scorefxn(test_pose)
    test_pose.dump_pdb(f"design_{i:03d}.pdb")
    print(f"Design {i}: {score:.2f}")

4. Protein-Protein Docking

#!/bin/bash
#SBATCH --job-name=rosetta_dock
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=12
#SBATCH --time=48:00:00

module purge
module load rosetta/3.15

# Docking protocol
docking_protocol.linuxgccrelease \
  -s complex.pdb \
  -partners A_B \
  -dock_pert 3 8 \
  -nstruct 1000 \
  -out:path:all docking_output/

🔧 Common Options

Input/Output

Option	Description
`-s <file>`	Input structure (PDB)
`-l <listfile>`	List of input structures
`-nstruct N`	Number of output structures
`-out:suffix <suffix>`	Add suffix to output files
`-out:prefix <prefix>`	Add prefix to output files
`-out:path:all <dir>`	Output directory

Database

Option	Description
`-database <path>`	Rosetta database path (auto-set by module)

Scoring

Option	Description
`-score:weights <file>`	Score function weights file
`-restore_pre_talaris_2013_behavior`	Use older score function

Performance

Option	Description
`-mute all`	Suppress most output
`-constant_seed`	Use constant random seed (reproducibility)
`-jran <seed>`	Set random seed

📚 Documentation & Tutorials

Online Resources

Official Documentation: https://docs.rosettacommons.org/
RosettaScripts: https://www.rosettacommons.org/docs/latest/scripting_documentation/RosettaScripts/RosettaScripts
PyRosetta Tutorials: https://www.pyrosetta.org/documentation
Rosetta Forums: https://www.rosettacommons.org/forum

Local Examples

# Rosetta demo files
ls $ROSETTA_ROOT/demos/

# PyRosetta tutorials
ls $ROSETTA_ROOT/source/src/python/PyRosetta/src/

Common Workflows

Protein Design
Load structure → Define designable positions → Run FastDesign → Analyze results
Structure Prediction
Generate fragments → Run AbinitioRelax → Cluster structures → Select models
Protein-Protein Docking
Prepare complex → Randomize orientation → Run docking → Score interface
Loop Modeling
Define loop region → Generate fragments → Build loops → Refine

🐞 Troubleshooting

Common Issues

1. "Cannot find database"

# Database path should be set automatically
echo $ROSETTA_DATABASE

# If not set:
module unload rosetta/3.15
module load rosetta/3.15

2. "Segmentation fault" or crashes - Check if you're on correct node architecture - AVX512 binaries will crash on AVX2-only CPUs

echo $ROSETTA_ARCH  # Verify loaded architecture
grep avx /proc/cpuinfo | head -1  # Check CPU capabilities

3. PyRosetta import fails

# Ensure miniforge3 is loaded first
module load miniforge3
module load rosetta/3.15

# Verify PYTHONPATH
echo $PYTHONPATH | grep rosetta

4. "Cannot find input PDB"

# Use absolute paths or ensure working directory is correct
rosetta_scripts.linuxgccrelease -s $PWD/input.pdb ...

5. Out of memory

# Request more memory in SLURM
#SBATCH --mem=16G

# Or reduce number of parallel operations

💡 Best Practices

Job Submission

Always use batch scripts - Don't run large jobs interactively
Request appropriate resources - Rosetta is CPU-intensive
Use -nstruct wisely - Generate enough decoys for sampling
Set random seeds for reproducibility with -constant_seed

Performance Tips

Use AVX512 nodes for compute-intensive jobs (faster)
Parallelize with -nstruct not MPI (unless protocol supports it)
Use silent files for large-scale runs (more efficient than PDB)
Profile first - Test with small -nstruct before full production

Data Management

Silent files - More compact than individual PDBs bash # Extract structures from silent file extract_pdbs.linuxgccrelease -in:file:silent output.out -in:file:tags tag1 tag2
Score files - Track energy terms bash # Rosetta creates score.sc automatically less score.sc
Checkpoint your work - Long jobs may timeout bash # Use multiple short jobs instead of one very long job

# For molecular visualization
module load pymol

# For trajectory analysis (if applicable)
module load vmd

# For sequence analysis
module load blast

📞 Support

Local Admin: uliano.guerrini@unimi.it Rosetta Forums: https://www.rosettacommons.org/forum Bug Reports: https://github.com/RosettaCommons/rosetta/issues

📝 Citation

If you use Rosetta in your research, please cite:

Leaver-Fay A, et al. (2011). ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol 487:545-574.

For PyRosetta:

Chaudhury S, et al. (2010). PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta. Bioinformatics 26(5):689-691.

Last Updated: October 12, 2025 Module Version: rosetta/3.15 Installation: /sw/rosetta/3.15_avx2 and /sw/rosetta/3.15_avx512