Bioinformatics Tools - Seoul National University · Bioinformatics Tools Byoung-Tak Zhang and Chul Joo Kang School of Computer Science and Engineering SeoulNationalUniversity (C)

Bioinformatics Tools

Byoung-Tak Zhang and Chul Joo KangSchool of Computer Science and Engineering

Seoul National University

(C) 2001 SNU CSE Artificial Intelligence Lab (SCAI) 2

Contents

1. Sequence Alignment

2. Multiple Sequence Alignment

3. Pattern Finding

4. Structure Prediction

5. DNA Microarray

6. Major Tools for Proteomics


Bioinformatics Tools

2D Gel, MALDI-TOFHardwares forProteomics

GeneX, GOE, MAT, GeNetDNA Microarray

Bend.it, RNA Draw, NNPREDICT, SWISS-MODEL

Structure Prediction

GRAIL, FGENEH, tRNAscan-SE, NNPP,eMOTIF, PROSITE, ChloroP

Pattern Finding

Clustal W, MacawMultiple SequenceAlignment

BLAST, FASTASequence Alignment

Tools or DatabaseProblems


Bioinformatics Based on Sequence

! Genome sequencing project: get huge sequencedata

! Rretrieving useful information from sequence data- Find genes and other elements

- Classification

- Predict its function

- Others

! Read the articles on Nature and Science


1. Sequence Alignment


Pairwise Sequence Alignment

! Basic job handling sequences! Alignment between two sequences! Sequence database search! Finding similar DNA or protein sequences

! Smith-Waterman algorithm! BLAST! FASTA


Smith-Waterman Algorithm (1)

! Finding local alignments

! Using dynamic programming

TCAT*G*CATTG


! Best results and slow performance

! Can grasp results missed by BLAST or FASTA.! Available on web:

spiral.genes.nig.ac.jp/homolgy/ssearch-e.shtml! Ref : Smith, T. F. and Watermann, M. S. (1981)

Identification of common molecular subsequence.Journal of Molecular Biology, 147: 196-197.

Smith-Waterman Algorithm (2)


BLAST & FASTA

! Using heuristic algorithms! Word based match! Faster than Smith &Waterman! BLAST: www.ncbi.nlm.nih.gov/BLAST

Ref: Altschul, S. F., et al. (1990) Basic local alignmentsearch tool. Journal of Molecular Biology, 215, 403-410.

! FASTA: www.ebi.ac.uk/fasta3Ref: Wilbur, W. J. and Lipman, D. J. (1983) Rapid

similarity searches of nucleic acid and protein databanks. Proceedings of the National Academy ofSciences USA 80, 726-30.


The BLAST Search

For the query, find the listof high scoring words oflength W.

Compares the word list tothe database and identifiesexact matches.

For each word match,extends the alignment inboth directions to findalignments that scoregreater than a threshold ofvalue S.


BLAST Result

Program: BLASTP

Query: human MYB

binding protein

Database: Swissprot


2. Multiple Sequence Alignment


Multiple Sequence Alignment

! Align multiple sequences! Finding well conserved regions

4Finding motifs or other elements

! Building gene families! Constructing a phylogenetic tree! Using whole dynamic programming ?

4O(n^k) problem (n = sequence length, k = number ofsequences)

! Clustal W! Macaw


Clustal W

! Step 1: Pairwise alignment between two sequences

! Step 2: Making sequence weights and clusters

! Step 3: Alignment between most similar sequences orclusters

! Step 4: Making one optimal output sequence

Ref: Higgins, D. G. and Sharp, P. H. (1988) CLUSTAL: Apackage for performing multiple sequence alignmenton a microcomputer. Gene 73, 237-244.


Multiple Alignment by Clustal W


Phylogenetic Tree Construction Basedon Clustal W Results


3. Pattern Finding


Pattern Finding

! DNA sequences have all information abouttheir products4Gene finding

4Other sequence element finding

4Motif finding

4Localization prediction

4And others


Gene Structure


Gene Finding

! From an unannotated DNA sequence, findputative expressive regions

! Distinguish exon and intron regions (ineukaryote).

! Solutions4Compare known genes (alignments)4Simply find conserved region4Using statistical models like hidden Markov

models and others


GRAIL: Find Exon and Intron

! In eukaryotes, RNA is processed.

! Intron: splice out

! Exon: join together

! GU – AG rule! GRAIL: exon prediction from a

genomic sequence withRepeatMasker filtering

! Neural networks which combine aseries of coding predictionalgorithms

! http://compbio.ornl.gov/Grail-1.3! Ref: Einstein et al. (1992) Computer-

based construction of gene usingthe GRAIL gene assembly program.Oak Ridge Natl. Lab. Report TM-12174.


GRAIL Results


FGENEH: Prediction of Multiple Genesin Human DNA Sequences! Finding genes from DNA sequences

! HMM based human gene structure prediction

! Available on web:

http://searchlauncher.bcm.tmc.edu/seq-search/gene-search.html

! Ref: Algorithm is trained on our data and based on HMMsimilar with Genescan (Burge, C. and Karlin, S. (J.Mol. Biol. 1997, 268, 78-94.) and Genie (Kulp, D.,Haussler, D., Reese, M.G., and Eeckman, F. H. (1996),Proc. Conf. on Intelligent Systems in MolecularBiology '96, 134-142).


FGENEH: Prediction of Multiple Genesin Human DNA Sequences


tRNAscan-SE: Find tRNA Genes

! Finding tRNA from genomic sequence

! Finding polII promoter sites, searching conservedsecondary structures - easier problem than finding proteincoding genes

! Combining several earlier programs and algorithms

! Http://www.genetics.wustl.edu/eddy/tRNAscan-SE/

! Ref: Lowe, T. M. and Eddy. S. R. (1997) tRNAscan-SE:A program for improved detection of transfer RNAgenes in genomic sequence. Nucl. Acids Res. 25: 955-964.


tRNAscan-SE Results


Sequence Element Finding

* Promoter region of E.coli lacZ operon

! Promoter and other elements: regulatory and other protein bindingsites


NNPP: Neural Network PromoterPrediction

! Finding promoter sequences on DNA sequences

! Time-delay neural networks

! Two feature layers4Recognizing TATA boxes

4Recognizing initiators

! http://www.fruitfly.org/seq_tools/promoter.html


Neural Network Promoter Prediction


Motif Finding

! Find conserved residues from DNA or protein sequences.

! Commonly, conserved residues are related to protein functions.

Motif of one class of zinc finger protein(C2H2) Cx{2,4}Cx{12}Hx{3,5}H

Solutions:• Finding consensus sequences

• Alignment

• Position specific weight matrix

• Hidden Markov models


Motif Databases: eMOTIF, PROSITE

! Prosite, eMotif: protein motif databases

! eMotif (motif.stanford.edu/emotif)

eMotif Maker: building new motif from multiple alignment

eMotif scan: find motif pattern

eMotif search: find motif from protein sequence

! Ref: Huang, J. Y. and Brutlag, D. L. (2001) The eMOTIF Database.Nucleic Acids Research, 29(1), 202-204

Craig et al. (1998) Highly specific protein sequence motif forgenomic analysis. Proc. Natl. Acad. Sci. 95: 5865-5871


eMOTIF (1)


eMOTIF (2)


eMOTIF (3)


PROSITE (1)

! PROSITE: Database of protein families anddomains (http://www.expasy.ch/prosite/)

! Consists of biologically significant sites, patternsand profiles that help to reliably identify to whichknown protein family (if any) a new sequencebelongs.


PROSITE (2)

e.g.) C2H2 zinc finger

protein


ScanPROSITE

! Compares query sequences

(protein) to Prosite.

! Find patterns


Others: Find Gene Destination

! After translation, proteins are located to proper places(nuclear, mitochondria, outside of cell…)

! That process determined by amino acid sequences onproteins4Finding signal peptide

4Predicting protein targets


ChloroP: Prediction of ChloroplastImporting Protein (1)

! Predicting chloroplast targeting transit peptides

! Predicting cleavage sites

! Neural network based method! Ref: Emanuelsson, O., Nilesen, H., and von

Heijne, G. (1999) A neural network-basedmethod for predicting chloroplast transitpeptides and their cleavage sites. ProteinScience 8, 978-984.

! Service site:http://www.cbs.dtu.dk/services/ChloroP/


ChloroP: Prediction of ChloroplastImporting Protein (2)


4. Structure Prediction


Structure Prediction

! DNA, RNA and protein structure prediction

! These structures affect their function, stability, etc.

! Protein structure prediction by calculation ofbiochemical properties of each amino acid4In most cases of protein secondary or tertiary structure

prediction, it is a terribly huge computing job (almostimpossible)

! Prediction based on known structures


DNA & RNA Structure

! DNA & RNA strands: dynamic one, not linear form

! DNA: regionally denatured (breathing), bending4Affects its expressions and others.

! RNA: forming intra strand pairing4Affects its stability, function (RNase), and others.


DNA Bending

! DNA structure is flexible.

! DNA double strand isbended by protein binding.

! The bending of the DNAstructure will influenceother proteins to bind to it.


Bend.it: Predict DNA Bendability

! Predicting DNA curvatures from DNA sequences

! The curvature is calculated as a vector sum ofdinucleotide geometries (roll, tilt and twist angles).

! Expressed as degrees per helical turn.! Ref: Munteanu, M. G., et al. (1998) Rod models of

DNA: sequence-dependent anisotropic elasticmodelling of local bending phenomena.Trends Biochem. Sci. 23(9), 341-346

! Service site:http://www2.icgeb.trieste.it/~dna/bend_it.html


RNA Draw: Predict RNA SecondaryStructure on PC (1)! Predicting RNA secondary conformation under given E

state

! Using dynamic programming, base pair probability and Eparameters

Clover leaf structure of tRNA and its 3D structure


E.coli tRNA-Ilepredictedstructure on 37 C

RNA Draw: Predict RNA SecondaryStructure on PC (2)


Protein Structure

! Protein function: determined by its structure(e.g.: enzyme inactivated by heat - because its structurewas changed by heat.)

Primary structure (amino acidsequence) determine itssecondary structure (partialstructure: α helix, β-sheet) andtertiary structure


Secondary Structure Prediction

! Prediction of secondarystructure4α-helix4β-sheet

! Can be helpful for its 3Dconformation study.

! Solutions4Calculate “propensity” for each

amino acid to be in helix, sheet…4Calculate “information content”4Neural network based method4Nearest neighbor


NNPREDICT: Protein SecondaryStructure Prediction! Two-layer, feed-forward neural network! Finding secondary structure from amino acid sequences

! http://www.cmpharm.ucsf.edu/~nomi/nnpredict.html


Tertiary Structure Prediction

! Using E state?: only select lowest E state - impossible

! Using homology modeling: similar sequence = similarstructure


SWISS-MODEL

! Predicting protein 3D structure


Swiss-Pdb Viewer


SWISS-MODEL

! Ref: Guex, N. and Peitsch, M. C. (1997) SWISS-MODEL and the Swiss-PdbViewer: Anenvironment for comparative protein modelling.Electrophoresis 18:2714-2723.Guex, N. and Peitsch, M. C. (1999) Molecularmodelling of proteins. Immunology News 6:132-134.

! Available on web:

http://www.expasy.ch/swissmod/SWISS-MODEL.html


5. DNA Microarray


DNA Microarray (1)

! Monitoring of gene expression

! cDNA chip: spotting cDNAs

! Oligonucleotide chip: spottingshort DNA sequences


! DNAs are spotted on glass surfaces.

! Applications: drug effects, drug metabolism, disease diagnosis, findinggene pathway, finding disease genes, etc.

DNA Microarray: cDNA chip (2)


! Short (10~25 nt) oligonucleotides on chip

! Applications: find sNPs, detect mutation, sequencing by chip,genotyping

DNA Microarray: oligonucleotide chip (3)


! Processing DNA chip images to obtain qualified data

! Data mining from DNA chip images like clustering (hierarchicalalgorithms, partitioning algorithms…)

DNA Microarray (4)


! Ref: Schena, M., et al. (1996) Parallel humangenome analysis: Microarray-based expressionmonitoring of 1000 genes. Proc Natl. Acad. Sci.USA. 93(20):10614-9.

! Useful websites:4www.affimatrix.com

4www.genechip.co.kr

4inkage.rockefeller.edu/wli/microarray

4cmgm.stanford.edu/~plf58/microArray

DNA Microarray (4)


6. Major Tools for Proteomics


Major Tools for Proteomics: 2D Gel &MALDI-TOF

! Proteomics:4Protein expression profile

4Protein interaction

4Protein structure

4Protein variation

! Why proteomics?4Post-Genomics Era: Understanding functions

4Applications in target discovery

4Novel proteins and genes

4Disease associated proteins

4Elucidate pathways and processes in disease/cell biology


2-D Gel Electrophoresis

! Separates proteins by their netcharge andmolecular weights.

! Displays whole proteins

! Like DNA chip, compares expression underdifferent treatments, and detects tissues or cellspecific expressions.4Finding specific proteins for further study

! Poor reproduction problem


2-D Gel Image of Human Liver


2-D Electrophoretic Gel Images


MALDI-TOF: Matrix-Assisted LaserDesorption Ionization! Detection of protein molecular weights (mass

spectrophotometer)

! Can determine protein sequence.


Other bioinformatic work

! Sequence assembly4From scattered DNA sequences, construct full sequence

! Specific database warehousing4Structural databases, protein block databases, drug databases

! sNPs4Find single nucleotide polymorphisms in population

! Construct gene to gene, protein to protein and gene toprotein interaction pathway

Documents

Bioinformatics Tools - Seoul National University · Bioinformatics Tools Byoung-Tak Zhang and Chul Joo Kang School of Computer Science and Engineering SeoulNationalUniversity (C)