Bioinformatics course

1
Homology vs. similarity

• Homology Evolutionary relation of two proteins
  with similar biological function that indicates
  common ancestry.
• Sequence similarity is observable homology is a
  hypothesis based on observation.
• We want to know whether two sequences are truly
  homologous because this will enable us to make
  conclusions about their probable structure and
  function.
• Sequence similarity can be global (overall
  sequence similarity) or local (motifs) in
  distinguishing probable homologues

2
Basics of Data Base Search

• How to find similar sequences (query sequence
  versus database)
• 1.      Use a similarity matrix (substitution
  matrix) to score the similarity of amino acids.
• 2.      Generate all possible alignments and
  calculate a score for each alignment
• 3.      The optimal alignment is the alignment
  with the highest score.
• Procedure of exhaustive enumerations and scoring
  of all alignments is not feasible.
•  

Number of all possible alignments for two
sequences of lengths n
3
Most widely used algorithms

• Two basic types of algorithms
• Needleman-Wunsch algorithm1,2
• Global algorithm which gives an overall best fit
  alignment of the entire sequence.
• rigorous algorithm to find optimal solution.
• requires tremendous amount of computing power.
• not sensitive for highly diverged sequences
• Smith-Waterman algorithm 3
• Local alignment procedure which tries to find a
  sub sequence (or several small) subsequences of
  high similarity.

Ref 1 Needleman, S.B. Wunsch, C.D. 1970. J.
Mol. Biol. 48, 443-453. 2 Gotoh, O. 1982. J.
Mol. Biol. 162, 705-708. 3 Smith, T.F.
Waterman, M.S. 1981. J. Mol. Biol. 147, 195-197.
4
How to derive score parameters for sequence
alignments

• Substitution matrices
• general idea
• from pairwise alignment of proteins derive
  probabilities pab for exchanging amino acid (or
  nucleotides) a and b and compare it to the
  expected probability in a random model R, pRqaqb
• Questions
• (a) How to score an alignment ?
• (b) What is a statistically valid set of protein
  sequences ?
• (c) What is a good random model ?
• (d) Different pairs of proteins have evolved
  from a common ancestor in a different amount.
• (e.g. compare homologs in a human and mouse to
  homologs in human and E.coli)
• (e) How to find the best alignment?

5
Examples of Substitution matrices

• Dayhoff PAM Matrices
• Dayhoff, Schwartz, Orcutt (1978). A model of
  evolutionary change in proteins. In Dayhoff ,
  Atlas of Protein sequence and Structure, Vol.5,
  NBRF, Washington, pp.345-352.
• BLOSUM matrices
• Henikoff, Henikoff (1992). Amino acid
  substitution matrices from protein blocks. PNAS
  89, 10915-10919.
• In both cases S(i,j) is determined (in a
  simplified way) by
•  
•  
• Nexch(i,j) are the occurrences substituting a.a.
  i with j Ni and NJ are the number of occurrences
  for the individual amino acids i and j

6
Parameters of the substitution matrices

• Both matrices have a parameter attached, these
  characterized the selection of proteins which are
  used to calculate the matrices
• PAM n
• point accepted mutation one PAM unit is
  equivalent to an average change of 1 of all
  amino acids
• first use closely related sequences, then expand
  using a model of evolution
• widely used PAM250
• BLOSUM m.
• use only ungapped, aligned regions of protein
  families (BLOCKS)
• the sequences from each block are clustered, two
  sequences in the same cluster if the sequence
  identity is large than a cutoff (m)
• smaller values of m means more diverse sequences
  
• BLOSUM 62 is widely used
•  

7
PAM point accepted mutation

• PAM scores are derived from alignments of
  closely relatedsequences, i.e., proteins whose
  function is known to be the same (Hemoglobin,
  cytochrome c, ribosomal proteins, RNase A...)
  from many organisms. The original PAM scoring
  matrix was derived by Margaret Dayhoff, a pioneer
  in sequence analysis
• Numbers may be expressed in terms of
  time-dependent probability matrices (P(t)) One
  PAM unit is the time required to achieve an
  average change of 1 in the amino acid positions.
  The original aim was to relate observed changes
  to the evolutionary distance between organisms,
  as reflected by the geological record. Thus PAM
  units may be expressed in millions of years of
  evolution.
• PAM250 will be drawn from a more diverse sequence
  alignment than PAM100.

8
PAM 250 Substitution matrix expressed as log odds
9
Similarity score is the sum of the matrix
elements in an alignment

• Residue pairs with scores above 0 replace each
  other more often in related sequences than in
  random sequences. This is an indication that both
  residues can carry out similar functions (similar
  size, hydrogen bonding, etc). A score exactly
  equal to zero indicates amino acid pairs that are
  found as alternatives at exactly the frequency
  predicted by chance. Residue pairs with scores
  less than 0 replace each other less often than in
  random sequences and might be an indication that
  these residues are not functionally equivalent.
• The score of an alignment is calculated as the
  sum of the substitution matrix elements in this
  alignment. This can be derived by assuming that
  all positions in a protein sequence are
  independent. Then the odds for the alignment is
  given by
• In practice one deals with sums rather than
  products.

10
Example

• Scoring the distance between two sequences with
  PAM250
• for example, two sequences from the EGF domain
  of rabbit and pig fertilin)
• QNCNN
• EKCHN
• S211222

11
BLOSUM the BLOcks SUbstitution Matrix.

• Scores are derived from alignments of distantly
  related sequences, without regard to function
• Should give a better substitution matrix for more
  distantly related sequences than the PAM
  matrices. Also, as PAM is limited to proteins of
  known function for its derivation, you have more
  sequences contributing to the BLOSUM numbers
  (better statistics)
• the sequence alignments are the from the BLOCKS
  database, with the numerical value derived from
  the cutoff value for the diversity of the
  sequence
• BLOSUM62 (sequences are gt62 identical) will be
  drawn from a less diverse sequence alignment than
  BLOSUM35 (where the sequences are gt35 identical)

12
Using the BLOSUM62 log odds matrix to score an
alignment add the numbers in the matrix
If you have a D above a W in an alignment, the
score is -3. For F to W, the score is 1.
Everytime F matches F, or D,D, add 6.
13
Other scoring matrices

• Gaston Gonnet and coworkers derived a matrix
  much like PAM250 by using pairwise alignments of
  all the sequences known in 1992, in an iterative
  fashion starting with alignments based on PAM250.
  They noted that their results were different when
  they used closely related sequence alignments vs.
  more distantly related ones.
• Identity matrix sort of the original, but only
  useful if it is scored according to the frequency
  of occurrence of amino acids in the database.

14
Concepts of protein structure prediction

• Why is there a need for protein structure
  prediction ?
• the sequence of a protein is easily available
• the determination of 3D structures is still a
  slow process
• energy based methods
• free energy of the protein in the native state
  is minimal
• Anfinson experiment
• ab initio structure prediction is still an
  unsolved problem
• holy grail of computational biology
• knowledge-based methods
• parameters are extracted from currently known 3D
  structures
• examples
• secondary structure prediction
• fold recognition methods (threading)
• knowledge based force field terms are added to
  free energy term

15
EXPLOSION OF GENOMIC DATA

• Gene sequences from genome projects far
  outnumber the experimentally determined 3D
  structures
• Prediction of 3D structures of proteins is a
  necessity

16
Tertiary structure prediction methods
Search for Similar Gene or Protein Sequences
Align sequences to find common motifs that
correlate with structure and functions
Predict the 3D Structure and Function of a New
Protein by
Ab initio Structure Prediction
Comparative Homology Modeling
Protein Fold Recognition (protein threading)
17
Tertiary Structure Prediction

• Ab initio modeling from sequence to 3D structure
• Two approaches
• purely energy based
• (2) prediction of secondary structure and short
  long range contacts
• (1) uses a force field (molecular mechanics)
  method and molecular dynamics (or other global
  minimization algorithms) to find the native state
  of a protein
• Molecular mechanics computes the molecular energy
  based on classical (Newtonian) mechanics and
  considers molecules as atoms bonded with elastic
  bonds
• Molecular Energy Bond Energy Angle Energy
  Torsion Energy Electrostatic Energy Hydrogen
  Bonds Energy Solvation Energy SS Bridge
  Energy

18
Energy terms (see handouts)

• Bond Energy (Bond length, Bond Angle)
• Torsion angle energy (torsion motions around
  bonds, improper dihedrals)
• Electrostatics
• Van der Waals
• Hydrogen bond

19

• (2) prediction of secondary structure and long
  range contacts
• Secondary structure prediction derive propensity
  values of residues from statistical analysis of
  residues in known secondary structure
• More sophisticated methods Neural Network,
  combined prediction from MSA and HMM
• Long range contacts
• Tree-determinant residues
• Motifs
• Correlated mutations

20
Comparative Homology Modeling

• 3D structures of proteins come in families and
  superfamilies
• E.g. SCOP http//scop.mrc-lmb.cam.ac.uk/scop/
• families sequence identities high (gt 35), same
  functional residues
• superfamilies similar 3D fold some common
  functional motifs
• No universal definition of superfamilies
• . folds similar 3D fold
• Rule of thumb if two proteins have an alignment
  with a sequence identity gt 30 they have the same
  fold.
• More sophisticated methods for fold recognition
  3D profiles or threading
• Steps
• - for a target sequence find a homologous PDB
  template structure,
• - make an optimum alignment between the target
  and template sequences,
• - generate the the tertiary structure of the
  target using the template geometry.

21
Additional considerations
What is the secondary structure? Is it homologous
to other protein sequences? Is it homologous to
other protein structures? What is the best
sequence alignment between your target protein
and homologous PDB structures? Examine the
regions of insertions and deletions. Are they
located in the loop regions? On the surface? Is
the region hydrophobic or hydrophilic? The PDB
template might have functional sites and
established motifs. Does your target sequence has
the same features? If disulphide bridges are
present in the PDB template, are cysteine
residues aligned?
22
Basics of Secondary Structure Prediction

• Propensity values for secondary structures of
  amino acids
• Statistical analysis for the occurence of amino
  acids in regular secondary structures of a
  database of representative proteins
• assignment of secondary structure,
• X denotes one of the 20 aa, XAla, Val,..
• naX number of aa X in a-helical regions
• Na number of all amino acids in a-helical
  regions
• NX number of aa X in the database Ntot total
  number of all amino acids in the data base

frequency of amino acid X in a-helical
regions average frequency of a-helices in all
proteins Propensity values
23
Methods for prediction

• Classic method (Chou Fasman, 1985)
• simplified rules
• separate amino acids into groups of helix
  (b-strand) formers and breakers
• search for clusters of formers (four h-former out
  of six contiguous residues three b-former out of
  five residues extend the segments in both
  dimensions until a tetrapeptide of breakers is
  found
• later improvements
• Garnier Osguthorpe Robson (GOR) method
• influence of residue at postion j on secondary
  structure in the neighborhood of the residue is
  included
• main effect is statistically found in the range
  j-8 lt i lt j8

24
Improvements of the methods
major improvements larger databases multiple
sequence alignments neural network
method consensus prediction Meta server
25
Secondary Structure Prediction Servers

• APSSP2 www.imtech.res.in/raghava/apssp2/
• Advanced Protein Secondary Structure Prediction
  Server, GPS Raghava, Bioinformatics Center,
  Chandigarh
• PSIPRED bioinf.cs.ucl.ac.uk/psipred/index.html
• The PSIPRED Protein Structure Prediction Server,
  D. T. Jones, Department of Computer Science,
  University College London, UK.
• PROF www.aber.ac.uk/phiwww/prof/
• University of Wales, Aberystwyth, Computational
  Biology Group.
• PredictProtein cubic.bioc.columbia.edu/predictpro
  tein/
• The PredictProtein server , B. Rost, Columbia
  University, NY.
• SAM-T02sec www.cse.ucsc.edu/research/compbio/HMM-
  apps/T02-query.html
• HMM methods, K. Karplus, UCSC
• JPRED www.compbio.dundee.ac.uk/www-jpred/
• A consensus method for protein secondary
  structure prediction
• G. Barton, University of Dundee