Pair-Wise Sequence Alignment Methods and Tools

1
Pair-Wise Sequence Alignment Methods and Tools
2
Pair-Wise Alignment Global and Local
AlignmentsPart I Fundament Concept
3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
Scoring Matrices

• Alignment of two nucleotide sequences is
  traditionally scored using values which may be
  looked up in a weighted scoring matrix.
• The matrices most frequently used for scoring
  alignments of amino acid and nucleotide sequences
  come from the PAM (Percent (point) Accepted
  Mutation)and BLOSUM (Blocks Substitution
  Matrices) families.
• PAM
• The PAM matrices are constructed so that the
  highest scoring alignments using PAMn will be
  within n PAM units of each other. PAM1 is the
  base PAM matrix. It is constructed so that
  maximal scores are produced for alignments with
  only 1 mutation (99 conservation).
• To construct the PAMn matrix from PAM1(M1), the
  following formula is used

10
PAM100
11
Scoring Matrices (contd.)

• BLOSUM
• BLOSUM matrices are based on local multiple
  alignments of more distantly related sequences.
  Unlike PAM matrices, BLOSUM matrices were created
  from real amino acid data.
• For the creation of BLOSUM matrices, a database
  of multiple alignments without gaps for short
  regions of related sequences was derived. Within
  each alignment, the sequences were clustered into
  groups of sequences similar at by some threshold
  percent value.
• Substitution frequencies for all pairs of amino
  acids were calculated between the groups, to
  create the Block Substitution Matrix for each
  cluster.
• The number associated with the matrix is the
  minimum percent of identity of the sequences in
  the block. For example BLOSUM50 means that the
  sequences in this block are at least 50
  identical.

12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
Not only one best path
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
FASTP
FASTA
24
FASTA http//www.ebi.ac.uk/Tools/fasta33/index.ht
ml
25
FastP and FastA (1)

• FastA is an algorithm that attempts to speed up
  string matching over the standard optimal
  alignment.
• String matching using dynamic programming run in
  quadratic time. FastA uses direct addressing or
  k-tuple preprocessing to cut down the dynamic
  programming search space significantly. This
  results in reduced search time at the expense of
  some sensitivity.
• The FastA algorithm is implemented in the
  following 6 stages
• Locate hot spots
• Find the 10 best regions in the matrix
• Score using a substitution matrix

26
FastP and FastA (2)

• Combine initial regions from different diagonals
• Optimal alignment
• Presentation
• Locating Hot spots
• FastA allows the specification of a parameter
  called ktup. The ktup sets the basis word length
  for the comparisons between the query string and
  a given string in the database. ktup values are
  typically six for DNA sequences and two for
  protein sequences. For the DNA case each word is
  represented as a base 4 number that is also the
  index into the table.
• The matching ktup-length substrings are referred
  to as hot spots. To locate the hot spots, FastA
  creates a dictionary of all possible words of
  length ktup that occurs in the query sequence.

27
FastP and FastA (3)

• Each entry contains the offsets where this
  particular combination of 6 letters occur in the
  query sequence. In this way, for each word in
  the searched string, only the dictionary need be
  consulted to determine if and where the word
  occurs in the query string.
• Finding the 10 Best Regions
• A region is a sequence of consecutive hot spots
  on the same diagonal. Spaces between the hot
  spots are permitted.
• FastA ranks regions by giving each hot spot a
  positive score. The intervening space between
  consecutive hot spots in is given a negative
  score. The larger the gap the more severe the
  penalty.
• The score of the diagonal run is the sum of the
  hot spots scores and the interspot penalties.

28
FastP and FastA (4)

• Scoring with Substitution Matrix
• FastA next applies a substitution matrix to the
  10 best regions found above. The substitution
  matrix may be an amino acid or nucleotide based.
  This step allows different matches to be weighted
  differently.
• The single best subalignment found after the
  application of the substitution matrix is termed
  init1.
• Combining Initial Regions from Different
  Diagonals
• In this step, FastA checks to see if any of the
  initial regions from different diagonals may be
  combined to form a new higher scoring region.
  The score for the combined regions is the sum of
  the scores of the contributing regions less a
  joining penalty for each join.
• The score for the highest scoring region after
  this step is termed initn.

29
FastP and FastA (5)

• This step can be implemented using directed
  weighted graphs where the vertices are the
  subalignments from the last stage.
• The maximum weight path gives the initn
  alignment.
• Optimal Alignment
• In addition to initn, FastA computes an
  alternative local alignment score opt. This
  score is obtained by considering a narrow
  diagonal band in the dynamic programming matrix,
  centered along the init1 diagonal.
• Presentation
• Finally the database is ordered by either the opt
  or initn scores and the highest ranking result
  sequences are run thorough a full Smith-Waterman
  alignment.

30
(No Transcript)
31
(No Transcript)
32
BLAST http//blast.ncbi.nlm.nih.gov/Blast.cgi
33
BLAST (1)

• The Basic Local Alignment Search Tool (BLAST)
  program uses a heuristic algorithm to search for
  local alignments of a query string on a BLAST
  formatted database.(directly approximates
  alignments that optimize a measure of local
  similarity, the maximal segment pair (MSP)
  score.)
• It is reported to run 100 times faster than a
  Smith-Waterman serial search. There is a
  different BLAST version for each of the
  combinations of query types and database types.

34
BLAST (2)

• The BLAST database consists of three files for
  every FastA file input.
• The first contains all of the sequence headers,
  textual information about the amino acid or
  nucleotide sequence.
• The second contains the compressed sequences (2
  bits for each nucleotide, 5 bits for each amino
  acid).
• The third file contains an index of the
  compressed sequences so that they can be matched
  with the corresponding headers.
• The program runs in 3 rounds.
• Database Scanning (table search or Finite state
  machine)
• Seed Growing
• Combining Alignments

35
BLAST (3)

• Database Scanning
• BLAST searches the database using sequential
  search for short words (k-tups) of length W (Word
  Size) in the query string which score higher than
  T (Neighborhood Word Score Threshold).
• In BLAST 1.4, W is usually 3 amino acids or 11
  nucleotides, and in BLAST 2.0 this is usually 2
  amino acids or 2 sets of 5 nucleotides that are
  not contiguous.
• Once the tuple size has been selected, scanning
  may be accomplished in either of two ways.
• The first maps all k-tups of length W into a
  unique integer.
• The second approach for scanning a database is to
  construct a deterministic finite automata (DFA)
  based on the query word to scan the database.
  This second approach was chosen since it saved on
  time and space. The list of successful words is
  called the neighborhood. This information is
  stored in an index for the next round.

36
BLAST (4)

• Seed Growing
• In this round the matches found in the first
  round are used as "seeds" to "grow" the alignment
  in both directions.
• Combining Alignments
• Now the program attempts to combine multiple
  alignments.

37
(No Transcript)
38
(No Transcript)
39
Pair-Wise Alignment Global and Local
AlignmentsPart II Advance Concept
40
(No Transcript)
41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
(No Transcript)
46
(No Transcript)
47
(No Transcript)
48
(No Transcript)
49
(No Transcript)
50
(No Transcript)
51
(No Transcript)
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
(No Transcript)
57
(No Transcript)
58
Three Sequence Alignment
59
Three-Sequence Alignment algorithm (1)

• The definitions proposed by X. Huang (SAC, 1994).

1-gap block gap-open penalty is q1 and
gap-extension penalty is r1
2-gap block gap-open penalty is q2 and
gap-extension penalty is r2
a triple of 1-gap block
a triple of 2-gap block
Example
There are three possible forms for 1-gap and 2-gap
triple
60
(No Transcript)
61
(No Transcript)
62
Three-Sequence Alignment algorithm (2)

• Applying Hirschberg algorithm

63
Open Problems

• Local three sequences alignment? (no solution)
• How to align two whole genome sequences?
• How to align other types data, as secondary or 3D
  structure data? (or mixed data)

64
Extended reading
65
(No Transcript)
66
(No Transcript)
67
SSEA http//protein.bio.unipd.it/ssea/
68
(No Transcript)
69
(No Transcript)
70
CE http//cl.sdsc.edu/ce.html
71
CE-MC http//pathway.rit.albany.edu/cemc/
72
(No Transcript)
73
PRALINE http//zeus.cs.vu.nl/programs/pralinewww/
74
Tools and Sequences

• BioEdit (http//www.mbio.ncsu.edu/bioedit/bioedit.
  html)
• MEGA (http//evolgen.biol.metro-u.ac.jp/MEGA/)
• Sequences
• U07611
• JN400599
• EF126963

75
BioEdit
76
(No Transcript)
77
(No Transcript)
78
MEGA
79
(No Transcript)