Genetic Algorithm in Protein Structure Comparison

1
Genetic Algorithm in Protein Structure Comparison

• Chen Jinxiu,Feng yuan,Lan Man, Li Haiquan,Li
  Quan, Ren li'an
  
• Laboratories for Information Technology

2
Outline

• Background Protein Structure Comparison and
  Genetic Algorithm
  
• Overview of GA in structure comparison
• Case Study Related Protein Comparison
• Conclusion

3
1 Background Intuition
4
Background - Problem

• Find the similarity of two structure or
  substructure, more conserved than sequence.
  
• Three kinds of problems
• Single Protein Comparison, find the invariant
  part
  
• Related Protein comparison, find the essential
  substructure of a family
  
• General comparison, find the structure for motif

5
Protein Structure Comparison

• Presentation of structure, usually a linear
  sequence.
  
• Score function, intermolecular and
  intra-molecular distance.
  
• Comparison algorithm superposition based, graph
  based, GA based.
  
• Post-processing check

6
Major issues in comparison

• Huge Information
• Huge Searching space, a lot of alignment method
• So the algorithms are very important

7
Genetic Algorithm

• Basic Philosophy, excellent parents produce
  outstanding children.
  
• Of course, it is not true. By checking, remove
  unexpected ones.
  
• Steps
• Population a set of hypothesis
• Remove poor ones
• Produce next generation by combine good
  hypothesis.

8
2 GA in Structure Comparison

• Motivation
• Protein structure have huge information,
  comparison have huge searching space.
  
• GA can get the optimum very fast.

9
Basic Issues in GA comparison

• Design the hypothesis, must be biological
  significant
  
• Design genetic operations, must be biological
  significant too.
  
• Score function to evaluate hypothesis
• Constraints and Stop criteria.

10
Case Related Protein Comparison

• Biological Background
• A family of protein must have some crucial
  common structures, which is by revolution.
  
• Main Idea
• Align the secondary structure first, which
  means a large region, then find the detail local
  alignment. GA is only used in SSE level.

11
Cases Intuition

• Representation A sequence of SSE.
• Intuitive Goal

12
Case GA Steps

• Hypothesis
• GA Operations
• Swap Exchange with the whole type
• Crossover Exchange inside a type of SSE
• Mutate Alignment changes inside a SSE.
• Hop Inside a Hypothesis

13
GA Operation-swap
14
GA Operation -crossover
15
Details alignment

• Atom alignment after SSE alignment is fixed.
• Still a lot of means.

16
Score Function

• Formula

17
Flow Review

• Start from a population
• Evaluate each hypothesis
• GA operations on good ones
• Next generation

18
Conclusion

• GA is suitable for protein structure comparison
• GA is an efficient approximate method to find the
  optimum alignment.
  
• Hypothesis and GA operations are the key issues.
• Current research in this area is limited.

19
Open Problems

• Apply GA into other comparison approach, such as
  graph based.
  
• Apply GA into multiple structure Comparison
• Theoretical Research on GA