The Genome Access Course Multiple Sequence Alignments

1
TheGenomeAccessCourseMultiple
SequenceAlignments
2
Pairwise vs. Multiple Alignment

• Aligning two sequences according to the
  Smith-Waterman alignment is straightfoward, but
  the difficulty of expanding the alignment to more
  sequences increases exponentially.

3
Methods for Aligning Multiple Sequences

• Dynamic Programming
• Progressive
• Iterative
• Genetic Algorithm
• Hidden Markov Models (HMM)

4
Dynamic Programming

• A technique for designing efficient algorithms
  for optimization problems
• Some specific properties of the problem are
  required for the dynamic programming technique to
  be applicable
• Applicable when a large search problem can be
  broken down into stages
• Trivial solutions to sub-problems contribute to
  the overall solution

5
Optimal Alignment

• For a given group of sequences, there is no
  single "correct" alignment, only an alignment
  that is "optimal" according to some set of
  calculations.
• Determining which alignment is best for a given
  set of sequences is an individual decision.

6
Progressive Methods

• Add new sequences one at a time to a pairwise
  alignment generated with dynamic programming
• Most closely related aligned first
• Modeled by an evolutionary tree
• Sensitive to intial alignments

7
Progressive PairwiseMethods

• Most of the available multiple alignment programs
  use some sort of incremental or progressive
  method that makes pairwise alignments, then adds
  new sequences one at a time to these
  aligned groups.
• This is an approximate method!

8
ClustalW

• Weighted Clustal
• Performs pairwise alignments of all sequences
• Produces a phylogenetic tree by neighbor-joining
  method
• Aligns sequences sequentially

9
Other Software

• MultAlin
• SAM
• HMMER
• PIMA
• treealign
• PAM

10
Commercial Software

• VectorNTI Suite AlignX
• GCG Pileup
• DNAStar

11
Genetic Algorithm

• Machine learning algorithm
• Simulates evolutionary changes in the sequences
• Alignments not necessarily optimal
• Seeks to increase initial msa score by simulating
  gap insertion and recombination

12
Markov Models
A Markov model is a probabilistic process over a
finite set, S1, ..., Sk, usually called its
states. Each state-transition generates a
character from the alphabet of the
process. State transitions are determined by a
transition probability matrix, which is
ordinarily independent of history.
13
Hidden Markov Models

• A Markov model in which the states are hidden.
• A given state in the sequence cannot be
  determined, but probabilities can be calculated.
• Can be designed such that biological relevance
  can be assigned to a state
• Originally applied to speech recognition

14
Uses of Hidden Markov Models

• Multiple sequence alignment
• Gene prediction
• Protein families (Profile HMM)
• Fold Recognition
• GpC island detection

15
DNA Sequence Transition Tables
Normal DNA
CpG Island
16
Building a Profile HMM

1. Construct a transition matrix for same-length
   sequences with no gaps
2. Correct for single insertions
3. Correct for variable length insertions
4. Correct for constant deletions
5. Correct for variable length deletions

17
Sequence Alignment HMM
Match State
Delete State
Insert State
18
MSA with HMMs

1. Construct a profile HMM
2. Find most likely path for each sequence
3. Sequence of matching/insert/delete states is the
   alignment

19
Editing MSAs

• CINEMA
• MACAW
• BOXSHADE
• PRETTYBOX

20
Databases Based on MSAs

• PROSITE
• FINGERPRINTS
• BLOCKS