Transcript Protein structural prediction methods

CZ5225 Methods in Computational Biology
Lecture 8: Protein Structure Prediction
Methods
. Chen Yu Zong
Tel: 6874-6877
Email: [email protected]
http://xin.cz3.nus.edu.sg
Room 07-24, level 7, SOC1, NUS
August 2004
Protein Structural Organization
Proteins are made from just 20 kinds of amino acids
2
Protein
Structural
Organization
Protein has four
levels of structural
organization
3
Protein Folding:
Sequence-Structure-Function Relationship
4
Protein Folding:
Sequence-Structure-Function Relationship
5
Measuring Structural Similarity:
The use of RMSD
6
Measuring Structural Similarity:
7
Measuring Structural Similarity:
8
Measuring Structural Similarity:
9
Protein Structure Prediction:
10
Protein Structure Prediction:
11
Protein Secondary Structure Prediction:
•
Secondary structure forms early in protein folding process.
•
Identification of secondary structural elements makes the topology of
protein structure more obvious—so that similar ones can be identified
in a topology database such as TOPS.
•
Prediction of the positions and lengths of secondary structure
elements can be used as a prelude to "docking" these secondary
structural elements against each other
•
Useful guide in the construction or refinement of primary structure
alignments, and to the correct correspondence between parts of two
proteins' respective tertiary structures.
•
Useful for making some kind of intelligent guess about the higher
order structure of your protein
12
Protein Secondary Structure Prediction:
Traditional methods: CF, GOR – Accuracy 60%
Recent improvements: Neural network, homologous sequences – Accuracy > 70%
References:
•
"Prediction of the secondary structure of proteins from their amino acid
sequence", P. Y. Chou, G. D. Fasman, 1978, Adv. Enzymolog. Relat. Areas Mol.
Biol., 47, 45-147.
•
"GOR method for predicting secondary structure from amino acid sequence", J.
Garnier, J.-F. Gibrat, B. Robson, 1996, Methods Enzymol., 266, 540-553.
•
"Analysis of the accuracy and implications simple methods for predicting the
secondary structure of globular proteins", J. Garnier, D. J. Osguthorpe, B.
Robson, 1978, J. Mol. Biol., 120, 45-147.
•
"Improvements in protein secondary structure prediction by an enhanced neural
network", Kneller, 1990, J. Mol. Biol., 214, 171-182
13
Protein Secondary Structure Prediction:
Software:
•
•
•
•
•
•
•
•
Zvelebil, M.J.J.M., Barton, G.J., Taylor, W.R. & Sternberg, M.J.E. (1987). Prediction of
Protein Secondary Structure and Active Sites Using the Alignment of Homologous
Sequences Journal of Molecular Biology, 195, 957-961. (ZPRED)
Rost, B. & Sander, C. (1993), Prediction of protein secondary structure at better than 70 %
Accuracy, Journal of Molecular Biology, 232, 584-599. PHD)
Salamov A.A. & Solovyev V.V. (1995), Prediction of protein secondary strurcture by
combining nearest-neighbor algorithms and multiply sequence alignments. Journal of
Molecular Biology, 247,1 (NNSSP)
Geourjon, C. & Deleage, G. (1994), SOPM : a self optimized prediction method for protein
secondary structure prediction. Protein Engineering, 7, 157-16. (SOPMA)
Solovyev V.V. & Salamov A.A. (1994) Predicting alpha-helix and beta-strand segments of
globular proteins. (1994) Computer Applications in the Biosciences,10,661-669. (SSP)
Wako, H. & Blundell, T. L. (1994), Use of amino-acid environment-dependent substitution
tables and conformational propensities in structure prediction from aligned sequences of
homologous proteins. 2. Secondary Structures, Journal of Molecular Biology, 238, 693-708.
Mehta, P., Heringa, J. & Argos, P. (1995), A simple and fast approach to prediction of protein
secondary structure from multiple aligned sequences with accuracy above 70 %. Protein
Science, 4, 2517-2525. (SSPRED)
King, R.D. & Sternberg, M.J.E. (1996) Identification and application of the concepts
important for accurate and reliable protein secondary structure prediction. Protein Sci,5,
2298-2310. (DSC).
14
Protein Secondary Structure Prediction:
Types of amino acids
Hydrophobic
Hydrophilic, Neutral
Hydrophilic, Acidic
Hydrophilic, Basic
15
Protein Secondary Structure Prediction:
Types of Secondary Structures:
Alpha helix and Beta- sheet
16
Protein Secondary Structure Prediction:
Secondary Structures: Favored Peptide Conformation
17
Protein Secondary Structure Prediction:
Secondary Structures:
Computation of structural propensity of a residue
• Data derived from
proteins of known
structure is used
to calculate
'propensities' for
each amino acid
type for adopting
helix, sheet or turn
18
Protein Secondary Structure Prediction:
Secondary Structures:
Computation of structural propensity of a residue
Three states: alpha helix, beta sheet, turn
19
Protein Secondary Structure Prediction:
Structural propensity of
amino acids
Each residue is assigned to
one of the three classes:
•
•
•
Forming residues – favor a structure
Indifferent residues
Breaking residues – stop the extension
of a structure
20
Protein Secondary Structure Prediction:
Position specific turn parameters
21
Protein Secondary Structure Prediction:
Chou and Fasman procedure
•
•
•
•
•
•
Find helical initiation regions
Extend helices until they reach tetrapeptide breakers
Find beta initiation regions
Extend until they reach tetrapeptide breakers
Find turns
Resolve conflicts between alpha and beta
Somewhat subjective … often have overlaps. Chou and Fasman suggest using additional information:
•
alpha-beta pattern, i.e. does this look like an b-a-b structure ???
•
end probabilities – Chou and Fasman in later papers also tabulated the preferences for the residues to
occur at the amino and carboxyl terminal ends of a and b structures.
These can be used to resolve overlaps
Chou and Fasman did not provide an explicit algorithm for this conflict resolution, relying on their expert
judgment. This meant that each person’s prediction could be different. Most people are not experts.
"Prediction of the secondary structure of proteins from their amino acid sequence",
P. Y. Chou, G. D. Fasman, 1978, Adv. Enzymolog. Relat. Areas Mol. Biol., 47, 45-147.
22
Protein Secondary Structure Prediction:
23
Homology Modeling:
24
Homology Modeling:
Reference:
• Sanchez R, Sali A. Advances in comparative proteinstructure modelling. Curr Opin Struct Biol. 1997
Apr;7(2):206-14.
• Krieger E, Nabuurs SB, Vriend G. Homology modeling.
Methods Biochem Anal. 2003;44:509-23
• Rodriguez R, Chinea G, Lopez N, Pons T, Vriend G.
Homology modeling, model and software evaluation:
three related resources. Bioinformatics. 1998;14(6):523-8
• Alexandrov NN, Luethy R. Alignment algorithm for
homology modeling and threading. Protein Sci. 1998
Feb;7(2):254-8
25
Homology Modeling:
Basic Idea:
• Similar sequence=> Similar structure
• Structure is conserved more than
sequence
• Structure of new protein derived using
existing protein structures as templates.
• Changes are compensated for locally.
26
Homology Modeling:
Twilight Zone: below 25% sequence homology
27
Homology Modeling:
• Similar sequence=> Similar structure
28
Homology Modeling:
Step One:
• Align sequence of your protein
(unknown) with that of candidate
template proteins (known)
29
Homology Modeling:
Step Two:
• Select template proteins based on
sequence similarity and minimize their Xray structures
• The whole sequence can be matched by
one or more templates
30
Homology Modeling:
Step Three:
• Combine the main chain of the template proteins and
fill-in gap sections to generate a complete main
chain model of your protein
• Gaps are filled-in by using short sequences from a
sequence linker library, the selected short
31
Homology Modeling:
Step Three:
• Combine the main chain of the template proteins and fill-in gap sections
to generate a complete main chain model of your protein
• Gaps are filled-in by using short sequences from a sequence linker
library, the selected short sequences need to be exchangeable to the
section of your original protein.
32
Homology Modeling:
• Step Four: Adding
side chains to the
main-chain model
based on the
sequence of your
protein:
– Mutate and add
33
Homology Modeling:
Step Five:
• Minimization and MD of the
homology model of your protein
H
p2
1
1
k r (r  req ) 2 
k (   eq ) 2 

2
2
bond  stretch
bond  anglebending
 2m  
atoms
vn
[1  cos( n   )]

2

bond  rotation
 [V (1  e
H bond
0
 [V (1  e
S bond
 a ( r  r0' ) 2
)  V0 ] 

nonbonded
 a ( r  r0' ) 2
)  V0 ] 
0
[
Aij
12
ij
r

Bij
6
ij
r

qi q j
 ij rij
]
34
Homology Modeling:
• Swiss-Model - an automated homology modeling server
developed at Glaxo Welcome Experimental Research
in Geneva. http://www.expasy.ch/swissmod/
• Closely linked to Swiss-PdbViewer, a tool for viewing
and manipulating protein structures and models.
• Likely take 24 hours to get results returned!
35
Homology Modeling:
How Swiss-model works?
•
•
•
•
•
1)
2)
3)
4)
5)
Search for suitable templates
Check sequence identity with target
Create ProModII jobs
Generate models with ProModII
Energy minimization with Gromos96
• First approach mode (regular)
• First approach mode (with user-defined template)
• Optimize mode
36
Homology Modeling:
How Swiss-model works?
Program
Database
Action
BLASTP2
ExNRL-3D
Find homologous sequences
of proteins with known structure.
Select all templates with sequence
identities above 25%.
Generate ProModII input files
Generate all models
Energy minimization of all models
SIM
--
-ProModII
Gromos96
-ExPDB
--
37
Threading Methods:
• Similar proteins at the sequence level may have very different
secondary structures. On the other hand, proteins very different at the
sequence level may have similar structures. Why? Because the
protein function is determined by its functional sites, which reside in
the cores not the loops.
• Therefore, researchers propose the inverse protein folding problem,
namely, fitting a known structure to a sequence.
• The problem of aligning a protein sequence to a given structural
model is known as protein threading.
• Given a protein whose structure is known, we derive a structural
model by replacing amino acids by place-holders, each is associated
with some basic properties such as an alpha-helix or beta-strand or
loop of the original amino acids.
38
Threading Methods:
References and software:
• Lemer C., Rooman, M. J. & Wodak, S. J. (1996), Protein Structure
Prediction By Threading Methods: Evaluation Of Current Techniques,
PROTEINS: Structure, Function and Genetics, 23, 337-355.
• Bryant, S. H. & Lawrence, C. E. (1993), An empirical energy function
for threading a protein sequence through the folding motif,
PROTEINS: Structure, Function and Genetics, 16, 92-112.
•
Alexandrov NN, Luethy R. Alignment algorithm for homology modeling and
threading. Protein Sci. 1998 Feb;7(2):254-8
• Jones, D.T., Taylor, W.R & Thornton, J.M (1992), A new approach to
protein fold recognition, Nature,358, 86-89. (THREADER).
39
Threading Methods:
• Threading methods take the amino acid sequence of an
uncharacterized protein structure, rapidly compute models based on
a large set of existing 3D structures.
• The algorithm then evaluates these models to determine how well the
unknown amino acid “fits” each template structure.
• All the threading models in the second to most recent CASP
competition produced accurate models in less than half of the cases.
• However, threading is more successful than homology modeling when
attempting to detect remote homologies that can’t be detected by
standard sequence alignment.
40
Threading Methods:
Protein Threading Model
• Input:
– A protein sequence A with n amino acids
– A structural model with m core segments Ci:
• (1) Each core segment Ci has length ci.
• (2) Core segments Ci and Cj are connected by loop Li, which has length
between li-min and li-max.
• (3) The local structural environment for each amino acid position, such as
chemical properties and spatial constraints.
– A score function to evaluate a given threading.
• Output:
– T = {t1, t2, ..., tm} of integers, where ti is the amino acid position in A that
occupies the first position in core segment Ci.
41
Threading Methods:
Protein Threading Model
• An algorithm: Branch and bound
• Spatial constraints:
1 + SUM (cj + lj-min) <= ti <= n + 1 - SUM (cj + lj-min)
j<i
j >= i
ti + ci + li-min <= ti+1 <= ti + ci + li-max
• A score function (second order, considering pairwise interaction):
f(T) = SUM g1(i,ti) + SUM g2(i,j,ti,tj)
i
j>i
• Algorithm testing: self-threading and using structural analogs.
42
Ab initio Methods:
H
p2
 2m  
bond  stretch
atoms
• ab initio means from the beginning.
1
1
k r (r  req ) 2 
k (   eq ) 2 

2
bond  anglebending 2
vn
[1  cos( n   )]

2

bond  rotation
 [V (1  e
H bond
0
 [V (1  e
S bond
 a ( r  r0' ) 2
)  V0 ] 

nonbonded
 a ( r  r0' ) 2
)  V0 ] 
0
[
Aij
rij12

Bij
rij6

qi q j
 ij rij
]
• Ab-initio algorithms attempt to predict structure
based on sequence information alone (i.e., no
emperical structural info is considered).
• Although many researchers are working in this vein,
it is a science in progress – sometimes marginally
successful, but very unreliable.
• Methods: MD and Simplified models
43
Ab initio Methods:
References:
H
p2
 2m  
atoms

bond  stretch
1
1
k r (r  req ) 2 
k (   eq ) 2 

2
bond  anglebending 2
vn
[1  cos( n   )]

2
bond  rotation
 [V (1  e
S bond
 a ( r  r0' ) 2
)  V0 ] 
0
Aij
Bij
•
 r r
Hardin C, Pogorelov TV, Luthey-Schulten Z. Ab initio protein
structure prediction. Curr Opin Struct Biol. 2002 Apr;12(2):17681. Review.
•
Srinivasan R, Rose GD. Ab initio prediction of protein structure
using LINUS. Proteins. 2002 Jun 1;47(4):489-95.
•
Bonneau R, Strauss CE, Rohl CA, Chivian D, Bradley P,
Malmstrom L, Robertson T, Baker D. De novo prediction of
three-dimensional structures for major protein families.
J Mol Biol. 2002 Sep 6;322(1):65-78.
•
•
[V0 (1  e  a ( r  r0 ) ) 2  V0 ] 
'
H bond
[
nonbonded
12
ij

6
ij

qi q j
 ij rij
]
Bystroff C, Shao Y. Fully automated ab initio protein structure
prediction using I-SITES, HMMSTR and ROSETTA.
Bioinformatics. 2002 Jul;18 Suppl 1:S54-61
44
Ab initio Methods:
H
p2
 2m  
bond  stretch
atoms
1
1
k r (r  req ) 2 
k (   eq ) 2 

2
bond  anglebending 2
vn
[1  cos( n   )]

2

) V ] 
 [V (1  e
LINUS as an example: Local Independently Nucleated Units of Structure
bond  rotation
 [V (1  e
H bond
 a ( r  r0' ) 2
S bond
 a ( r  r0' ) 2
0
)  V0 ] 

nonbonded
0
[
0
Aij
rij12

Bij
rij6

qi q j
 ij rij
]
•
50 amino acids are folded at a time, in an overlapping fashion: 1-50, 2675, ...
•
Based on the idea that actual proteins fold by forming local secondary
structure first.
•
Side chains are simplified. Only 3 interactions are used:
– 1 repulsive: steric
– 2 attractive: H-bonds and hydrophobic
– Then the calculation of all possibilities for the search of the lowest free
energy
45
CZ5225 Methods in Computational Biology
Assignment 2
Option 1:
• Write a code for protein secondary structure prediction.
• Test your code on several selected proteins and compare your prediction results with those
from the PHD software at http://npsa-pbil.ibcp.fr
Option 2:
• Write a code for protein homology modeling
• Test your code on several selected proteins, compute the rmsd of each of your predicted
structures against an x-ray structure of that protein.
Option 3:
• Write a code for structural comparison of two structures of unequal number of atoms. Test
your code on several pairs of molecules/proteins and compute the rmsd between each
pairs
Requirement: Write a report about the theory, algorithm, testing results, and suggested
Improvement/future work and submit together with a soft copy of your code.
46