Rosetta

1
Rosetta

• Steven Bitner

2
Objectives

• Introduction
• How Rosetta works
• How to get it
• How to install/use it

3
Introduction

• Developed in the David Baker lab at University of
  Washington
• Winner of CASP (Critical Assessment of Structure
  Prediction) competition at Lawrence Livermore
  Labs, CA
• Implies that Rosetta is the best de novo
  predictor
• Rosetta is a protein prediction and docking
  software package
• Also used to design proteins from (nearly)
  arbitrary 3-D shapes
• November, 2003, Top-7 first synthetic protein
• Rosetta home
• Human Proteome Folding Project
• Also called the World Community Grid

4
TOP-7 protein designed, or synthesized, using
Rosetta
5
How Rosetta Works

• Minimize energy in the folded state
• Uses a combination of energy formulas based on
  the likelihood of particular structures, and the
  fitness of the sequence
• Side-chains simplified to a centroid located at
  center of mass of the side-chain
• Average of observed side-chain centroids in known
  structures
• Local sequence does not decide the local
  structure, it only biases the decision
• Non-local favorable conditions
• Buried hydrophobic fragments
• Paired ß strands
• Specific side-chain interactions

6
How Rosetta Works (Energy functions)
7
How Rosetta Works (Energy functions) cont.
8
How Rosetta Works cont.

• Side-chains are added using Monte Carlo methods
• Overlaps of side-chain centroids and backbone
  atoms are penalized
• Uses probabilistic ß-strand pairing and ß-sheet
  patterns
• Fragment Insertion - more later
• Fragment Assembly - more later

9
Fragment Insertion

• Finds three and nine residue fragments from known
  library and replaces unknown torsion angles with
  the known ones
• Scores all windows of three and nine residues
• Create fragment list with the 200 best three
  residue and 200 best nine residue fragments

10
Fragment Assembly

• Randomly choose a nine residue fragment from the
  top 25 fragments in the ranked list
• Score this replacement, negatives are kept
• Each simulation chooses a different random start
  and attempts 28,000 nine residue insertions
• Next 8,000 attempted three residue insertions are
  scored with the overall structure

11
Why its Fast

• Changes multiple angles simultaneously by using
  fragments from the library
• Angular changes are discrete, not continuous

12
Getting the Software

• Go to the website (bakerlab.org)
• Register by clicking on Rosetta Licensing
  Information
• Go to link in email that is sent to you
• Download

13
Installing software

• Upload onto a Linux machine, or other supported
  platform
• (see README_platform)
• UTDs Apache server does not work
• Unpack using tar zxvf filename
• Go into rosetta directory
• Make gcc
• Takes about 20 minutes
• This is the standard version

14
Different install versions

• Other ways to install than make gcc
• See the README in rosetta directory
• GCCDEBUG for use if you plan on making updates
  to the software

15
Using the downloaded software

• PDB file must be in the same directory as the
  program or the paths.txt file must be updated
• paths.txt must be updated for the data source the
  default is a non-existant directory
• User guide assumes a good knowledge of the
  system

16
Using Rosetta

• Rosetta on-line Server 200 residues at a time
  http//rosettadesign.med.unc.edu/index.html
• Robetta site down until mid October 06
• http//robetta.bakerlab.org/
• Downloaded software
• Can use res files to specify portions of the
  backbone, or you can select the residues that you
  wish to pack on the web server

17
Interpreting results

• Output file fields
• Rosetta Commons site also has similar document
  except the energy labels use E for energy in
  stead of LJ and LK for Lennard-Jones and
  Lazaridis-Karplus respectively as the prefix
• E.g. Lennard-Jones attractive score is Eatr in
  the Rosetta commons output file and LJatr in the
  Rosetta Design output file

18
Input PDB file 1HOE displayed using PyMol
19
Output PDB file 1HOE displayed using PyMol
20
1 u b qBefore ? After
21
Software used for this presentation

• Rosetta release 2.1.0
• Rosetta Design Server http//rosettadesign.med.unc
  .edu/index.html
• PyMol for visualization
• RCSB PDB http//www.rcsb.org/pdb/Welcome.do

22
References

• Rosetta Design Web Server http//rosettadesign.med
  .unc.edu/documentation.html
• Protein Structure Prediction Using Rosetta,
  Numerical Computer Methods, C.A. Rohl, C.E.
  Strauss, K.M. Misura, D. Baker, pp. 66-93, 2004
• README documentation included with rosetta2.0.1
• Rosetta Website https//www.rosettacommons.org/
• David Baker Lab Homepage http//www.bakerlab.org/