Homology Modelling

1

• Tutorial
• Homology Modelling

2
A Brief Introduction to Homology Modeling
3
Sequence-Structure-Function Relationships

• Proteins of similar sequences fold into similar
  structures and perform similar biological
  functions.
• The protein sequence has the intrinsic
  information to encode the protein structure.

4
The Noble Prize in Chemistry 1972

• Christian B Anfinsen
• "for his work on ribonuclease, especially
  concerning the connection between the amino acid
  sequence and the biologically active
  conformation"

5
The protein sequence is sufficient to specify its
3D structure

• From Nobel Lecture, December 11, 1972, by
  Christian Anfinsen

6
Sequence-gtStructure-gtFunction

• Widespread Automated DNA sequencing gt more
  sequence data than structure data
• Semi-Automated pipeline of structure
  determination is still not widespread.
• Nevertheless, structure is more conserved than
  sequence.
• Sequence homologs gt structural homologs
• See Chapter 9, Baxevanis and Ouellette 3rd edn.

7
Protein Structure Prediction vs Experimental
Determination

• From Chapter 9, Bryan Bergeron, Bioinformatics
  Computing, 2003 Pearson Education, Inc.

8
Structure Prediction from sequence

1. Homology (or comparative) modelling
2. Threading
3. Ab initio calculationsHomology modelling is
   most accurate and powerful

9
What is Homology Modeling?

• Homology modeling also known as comparative
  modeling uses homologous sequences with known 3D
  structures for the modelling and prediction of
  the structure of a target sequence.
• Homology modeling is one of the most best
  performing prediction methods that gives
  accurate predicted models.

10
How is Homology Modeling done

• Multistep process involves many steps such as
• Sequence alignment of target/query/unknown
  protein sequence to homologous sequence with a
  known structure
• structure modification of backbone
• side chain replacements
• Energy minimisation for refinement of structural
  model
• Validation of model with visual inspection and etc

11
Why Homology Modeling?

• The number of protein structures solved so far
  are fewer than the number of genes known.
• Proteins of biological interest with their
  orthologous proteins solved by X-ray
  crystallography or NMR can be modeled.
• Homology modeling is an important method used to
  predict the structures of membrane proteins, ion
  channels, transporters that are large and
  difficult to crystallize.
• Examples GPCR (G Protein-coupled receptor),
  cytochrome P450 etc.

12
Overview of the process of Homology Modeling

• A target sequence (the structure to be predicted)
• Identify the homologous sequence with known 3D as
  template
• Using homology modeling software such as Modeller
  for structure prediction (from the Sali Lab)
• Model evaluation and refinement

13
Pre-Modeling Stage Template Identification

• Target sequence in FASTA format as input
• Blastp against PDB
• Identify proteins with good hit
• Pairwise or multiple sequence alignment
• Further editing the alignment results
• Realign and identify the good structural
  template

14
Pre-Modeling Stage Preparing the Input Files
for Modeller

• PDB files for structural templates is required
• The PIR file from the alignment results
• The script file model.top to execute the Modeller
  program(latest versions use Python scripts)

15
In the Heart of Modeller

• From the Modeller manual

16
Evaluation of Predicted ModelGarbage in-Garbage
out

• The predicted model can be superimposed with
  known structure determined by experiment
• http//wishart.biology.ualberta.ca/SuperPose/
• The predicted model is normally evaluated by root
  mean square deviation (RMSD)

17

• From http//swissmodel.expasy.org//course/text/cha
  pter6.htm

18
Calculating RMSD

• N number of atoms, d the distance in Angstrom
  between corresponding atoms in the experimental
  and predicted protein structures.
• From Chapter 9, Bryan Bergeron, Bioinformatics
  Computing, 2003 Pearson Education, Inc.

19

• Some Rule of Thumb for Structural Modelling
• Proteins that share 35 to 50 sequence identity
  with their templates, will generally deviate by
  1.0 to 1.5 Å from their experimental counter
  parts.
• Crystallographic structures of identical proteins
  can vary not only because of experimental errors
  and differences in data collection conditions and
  refinement, but also because of different crystal
  lattice contacts and the presence or absence of
  ligands.

20
Quality of Model

• The correctness of a model is essentially
  determined by the quality of the sequence
  alignment used to identify the template.
• If the sequence alignment is wrong in some
  regions, then the spatial arrangement of the
  residues in this portion of the model will be
  incorrect.

21
Viewing the Model

• The predicted model is saved in PDB format that
  can be viewed by molecular visualizing software
  such as Rasmol, PyMol, MolMol, Sybyl etc.
• Viewing is an essential step to validate the
  quality of the predicted model.
• In this practical, Rasmol is used to view the
  predicted structure.

22
Model Refinement

• Gaps in sequence alignment represent
  insertion/deletion regions of target. Loop
  modeling is used to refine these regions (not
  cover in this practical)
• The predicted model can be further refined by
  energy minimization to remove unfavourable
  non-bonded contacts with force fields such as
  CHARMM, AMBER or GROMOS etc (not covered in this
  practical)

23
Web-Based Homology Modeling The SWISS-MODEL
Server

• The aim of the Internet-based SWISS-MODEL server
  is to provide a comparative protein modelling
  tool independent from expensive computer hardware
  and software.
• http//www.expasy.ch/swissmod/SWISS-MODEL.html
  

24
Steps involved in SwissModel http//swissmodel.exp
asy.org/

1. Take target sequence of unknown structure
2. Using BLAST to select closest homolog with known
   structure as structural template
   http//swissmodel.expasy.org/SM_Blast.html
3. Insert target sequence and homologous sequence to
   Web service http//swissmodel.expasy.org/SM_FIRST.
   html
4. Results will be emailed back to you.
5. Warning Structure needs to be analysed and
   validated

25
Simple Homology Modelling using Modeller

1. Take target sequence of unknown structure
2. Using BLAST to select closest homolog with known
   structure.
3. Using Clustalx or Jalview to do pairwise
   alignment between target sequence and structural
   homolog and manual adjustment
4. Inspection of missing structural features in
   structural homolog
5. Preparation of alignment file align.pir
6. Use Modeller7v7 software (http//salilab.org/model
   ler/) to do the homology modelling

26
Structure Validation

• Visual inspection
• Minimise torsion angles in disallowed regions of
  Ramachandran plots
• Maximised hydrogen bonding
• Minimised exposed hydrophobic residues
• Packing etc.
• Analysis e.g. run Procheck (http//www.biochem.
  ucl.ac.uk/roman/procheck/procheck.html), VADAR,
  Verify3D etc