Heuristic approaches

1
Heuristic approaches scoring matrices

• M.Prasad Naidu
• MSc Medical Biochemistry, Ph.D,.

2
Introduction

• Two algorithms are there in these methods
• BLAST
• FASTA
• FastA is an algorithm developed by Pearson and
  Lipman. Its more sensitive than Blast.
• Blast is an algorithm developed by Altschul et
  al., in 1990. It provides tools for high scoring
  local alignment between two sequences. Now a
  days, a gapped versions are available.

3
BLASTP algorithm

• Blast Algorithm involves the following steps.
• Breaking of the sequence into defined word size.
• Finding a match or HSP (High Scoring Pair).
• Alignment of the word and extending the alignment.

4
Breaking of the sequence into defined word size

• Query AILDTGATGDA
• Word size 4
• AILDTGATGDA

AILD ILDT LDTG DTGA TGAT GATG
ATGD TGDA
5
Finding a High scoring Pair

• MQVWGWAILDTVATDAAMLL
• AILD

6
Extending the alignment

• MQVWGWAILDTVATDAAMLL
• ..AILDTGATGDA

Parameters in BLAST result Percentage of
Homology Scoring of the alignment No of residues
aligned E-value
7
FastA algorithm

• The word size in FastA algorithm is defined as
  K-tuple.
• Generally the K-tuple for the algorithm is either
  3 or 4 for nucleotide sequences and 1 or 2 for
  protein sequences.
• FastA algorithm also involves the steps similar
  to that of the BLAST tool. But the alignment
  generation procedure is different.

8
Breaking of the sequence into defined k-tuple

• F A M L G F I K Y L P G C M
• 1 2 3 4 5 6 7 8 9 10 11 12 13 14

A B C D E F G H I K L M
2 13 1 5 7 8 4 3
6 12 10 14
N P Q R S T V W Y Z
11 9

9
A B C D E F G H I K L M
2 13 1 5 7 8 4 3
6 12 10 14
N P Q R S T V W Y Z
11 9

T 1 G 2 F 3 I 4 K 5 Y 6 L 7 P 8 G 9 A 10 C 11 T 12
3 -2 3 3 3 -3 3 -4 -8 2
10 3 3 3
The most occuring number in the algorithm is 3,
so the alignment starts after leaving three
characters or residues
10
Alignment of the sequences

• F A M L G F I K Y L P G C M
• T G F I K Y L P G A C T

Parameters in FASTA result Percentage of
Homology Scoring of the alignment No of residues
aligned P-Score
11
Scoring schemes

• Identity scoring matrix
• Residue to residue scores are represented here in
  the form of similarity.
• A 4 X 4 matrix is built for the nucleotides and
  20 X 20 matrix for the amino acids.
• For match score is 1 and mismatch is -1

A T G C
A 1 0 0 0
T 0 1 0 0
G 0 0 1 0
C 0 0 0 1
12
PAM Matrices

• These were first developed by Margaret Dayhoff
  and co-workers in 1978.
• This model assumes that evolutionary changes
  follow the markov model i.e. residual changes
  occur independent on the previous mutation. One
  PAM is a unit of evolutionary divergence in which
  there is 1 amino acid change but it doesnt
  imply that 100 PAM results in different
  aminoacids.
• Dayhoff and coworkers have calculated the
  frequencies of accepted mutations for 1PAM by
  analyzing closely related families of sequences.
• The scores are represented as log odd ratios.
• The 1PAM can be extended to any no of PAMS. For
  example, 1PAM table is extended to N X 1PAM.
• For closely related protein sequences, lower
  distance PAM is used and higher PAM is used for
  variying proteins.
• PAM 30 is used for closer proteins and PAM 250
  for divergent ones.

13
PAM 250 scoring matrix
14
BLOSUM Matrices

• These matrices are developed by Heinkoff and
  Heinkoff in 1991.
• The matrices have been constructed in a similar
  fashion as PAM matrices.
• The data was derived for local alignment of
  distantly related proteins deposited in the
  BLOCKS database.
• BLOSUM 30 is used for comparing highly divergent
  sequences and BLOSUM 90 is used for closely
  related proteins.
• Commonly used BLOSUM matrix is BLOSUM 62 that is
  used for proteins with 62 identities.

15
BLOSUM 62 Matrix
16
THANK YOU