Transcript No Slide Title

Interactive tools and
programming environments
for sequence analysis
TATACATAAAGACCCAAATGGAACTGTTCTAGA
TGATACACTAGCATTAAGAGAAAAATTCGAAGA
ATCAGTCGATAAATACAAACTTCATTTTACTGG
ATTAATCGCTGACAAAATTGCAAAAGAAAAAC
TGAATACTTACGTCCTCACTTATAAAAAAGCAG
ACGAAGCTATGCCTGCAGACGAAGCTATGCCA
ACTGATGTACCTAGTACTTCTGTTACTGGATCAA
CAATGGCAAAC………………….
Bernardo Barbiellini
Northeastern University
Overview
•
•
•
•
•
Matlab and Darwin – bioinformatics tools
Dotplot and Statistical signifance of alignments
Scoring Matrices from Evolution Model
Evolutionary Distances and Phylogenetic Trees.
Unified approach for the sequence alignment
and structure prediction
Matlab toolbox and Darwin
•
•
•
•
•
Computer language appropriate for bioinformatics
A workbench to automate repetitive tasks
Based on Linear Algebra & Statistics
Matlab toolbox developed by Mathworks
Darwin developed by Gaston Gonnet (ETHZ)
Extra features
•
•
•
•
Loading of and retrieval in sequence databases
Fast searching for sequence fragments
Sequence alignment
Generation of random sequences, distributions and
mutations
• Creation of Phylogenetic trees
• Plotting functions - matrix and vector arithmetic
• I/O comunicate with other programs
Calling Bioperl functions in MATLAB
Documentation by Brian Madsen (NU and coop at the Mathworks)
>> help perl
PERL calls perl script using appropriate operating system
PERL(PERLFILE) calls perl script specified by the file PERLFILE
using appropriate perl executable.
PERL(PERLFILE,ARG1,ARG2,...) passes the arguments ARG1,ARG2,...
to the perl script file PERLFILE, and calls it by using appropriate
perl executable.
RESULT=PERL(...) outputs the result of attempted perl call.
Visual Tool: Dotplot (1)
Pairwise sequence comparison
Visual Tool: Dotplot (2)
Filtered Image
The best alignment is achieved with dynamic programming . A score is obtained
Quantitative Tools To Check
Statistical Significance
extreme value distribution.
Score in bits
Simulation with random sequences
PAM Evolution Model
The score of a paiwise alignment is obtained by
using a scoring matrix.
We need a model to build scoring matrices.
This model is based on evolution in order to
calculate evolution distances between species.
PAM means Accepted Point Mutation
Step1: Order of the Amino-Acids
Step 2: Mutation Matrices
Markov Model pamX=(pam1)^X Stochastic matrices
Step 3: Distribution of
Amino Acids
Eigenvector of the mutation matrix (eigenvalue 1)
Step 4: Evolutionary time vs.
sequences differences
Step 5: Scoring Matrix
The Dayhoff scoring matrix is symmetric
Tree Construction 1:
Evolutionary distance calculations
Maximum Likelihood
Tree Construction 2:
Table of distances
PAM
Spinach
Rice
Mosquito
Monkey
Human
Spinach
Rice Mosquito Monkey Human
0.0 84.9
84.9
0.0
105.6 117.8
90.8 122.4
105.6
117.8
0.0
84.7
90.8
122.4
84.7
0.0
86.3
122.6
80.8
3.3
86.3 122.6
80.8
3.3
0.0
Tree Construction 3:
Neighbor joining algorithm
Unified approach for the sequence alignment
and structure prediction
Optimization
with Dynamic
Programming
approach
Protein
Protein
Protein
Structure
Needleman-Wunsch
Algorithm
or
Smith-Waterman
Algorithm
Viterbi Algorithm
HMM
Query
Protein
Protein
Subject
Protein (letter of
amino acids)
Structure (, , coil)
Scoring
Matrix
Log (Aij/pi)
Log (P(im)/pi)
Penalties
Gaps
Transition from structure
to another
Conclusions
• The highly efficient dynamic programming algorithms,
used in this integrated environment, are particularly suitable
for the high performance computers.
• Trees constructed using optimal PAM distances are better
than the routinesingle distance scores obtained using a
single scoring matrix.
• The unified approach for the sequence alignment and
structure prediction provides a powerful formalism for
biologists.
ASCC Northeastern University
Northeastern University (NU)/Hewlett-Packard (HP) Company
Collaborative Research Program on Bioinformatics
Bernardo Barbiellini, Assoc. Director, ASCC
Arun Bansil, Professor of Physics & Director ASCC.
Bill Detrich, Prof. Biochem. & Marine Biology, Director Bioinformatics M. S.
Kostia Bergman, Prof. Biology
Mike Malioutov, Stone Professor of Applied Statistics
Mary Jo Ondrechen, Professor of Chemistry
Nagarajan Sankrithi, graduate student NU
Imtiaz Khan, graduate student NU
Alper Uzun, graduate student NU
Larry Weissman, staff HP/Compaq
Barry Latham, staff HP/Compaq
Bob Morgan, staff HP/Compaq
Other Bioinformatics activities at ASCC
• BIO3580: DNA and Protein Sequence Analysis (2001,
2002)
• MATLAB BIOINFORMATICS TOOL presentation
(Robert Henson)
• Summer Institute of Mathematical Studies on
Bioinformatics (2002) (Professor Mike Malioutov)
• Student projects proposed by Dr. Matteo Pellegrini,
(Proteinpathways/UCLA).