A Splice Site Locating

1
A Splice Site Locating Analyzing Tool

• December 6, 2002
• BioE 142
• Kelly Han
• Diana Wong
• Tyler Hillman

2
Background

• Locations of introns and exons are valuable
• Splice sites are mysterious and hard to predict

3
Existing Research

• Gene prediction poor accuracy levels are
• 80 nucleotide level
• 45 exon level
• 20 whole gene level
• Many (15) papers on gene prediction
• GenScan using HMM
• GeneSplicer a statistical method that integrates
  multiple sources of known evidence
• FirstEF predict CpG-related and non-CpG_related
  first exons
• Hexon linear discriminant analysis
• MZEF, FGENES, etc.list goes on

4
Problems

• Accuracy of gene prediction may slowly improve,
  but basic algorithms behind these various
  approaches have changed little since 1997
• Most existing approaches look for a specific
  pattern or unique featuredoes not allow for
  generalization
• Coupling algorithmic method with human heuristics

5
Project Overview

• Part I
• Created a useful tool for users to visualize
  possible splice sites by color coding nucleotides
• Included basic analysis pattern ranking (freq)
• Allows the user to make educated decisions
• Part II
• Proposed experimental design
• Stand alone, theoretical approach

6
Part I

• Found all splice sites from an annotated genome,
  Drosophila melanogaster
• 6 bases up/down stream
• Statistically analyzed splice site patterns
• Ranked them
• Use data to find possible splice sites in a
  sequence inputted by user
• Be able to analyze any annotated genome the user
  wishes to input

7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
Entering search query
11
(No Transcript)
12
(No Transcript)
13
Part II

• Available Data Drosophila melanogaster gene
  sequences with annotated intron/exon boundaries
  determined by molecular biology bench work
• Our Question
• Are the sequences around the splice sites just
  RANDOM sequences? In other words, is there any
  uniqueness?

14
Our Approach--- Sequence Reconstruction Method

• Step 1 Cut the gene at known intron/exon
  boundaries
• Step 2 Reconstruct it using overlaps
• Step 3 Cut the same gene with a random pattern
• Step 4 Reconstruct it
• Compare result of Step 2 result of Step 4

15
Reconstruction--- By Graph Theory

• Reconstruction Method Hamilton Path
• Each vertex corresponds to a specific k-tuple
• Arcs map the overlap among the k-tuples
• If a k-1 nucleotide overlap is the suffix (last
  k-1 nucleotides) of a k-tuple t1, and it is also
  the prefix (first k-1 nucleotide) of a k-tuple
  t2, then an arc is drawn from t1 to t2.
• All the vertices of the graph must be visited,
  therefore mathematically sequence reconstruction
  Hamilton Path problem

16
Interpretation of Results

• / inconclusive
• -/- inconclusive
• /- significant
• The pattern at the boundary is more
  organized/unique than randomly selected sequence
• (Possible errors poorly chosen boundary
  sequence, reconstruction errors, etc.)