Bioinformatics Datamining Bacterial Genome Sequences

1
BioinformaticsData-mining Bacterial Genome
Sequences

• Professor Mark Pallen
• University of Birmingham

2
BioinformaticsDefinitions

• Fusion of Biology, Computer science, Mathematics
• Broad Meaning
• Any computationally intensive research with
  biological relevance
• Narrow meaning
• Computer-based analysis and archiving of
  macromolecular sequence data

3
The Scope of Bioinformatics
Algorithm Development

• Domain Hunting
• Detailed Analysis by Expert

• Interface Design
• Graphics
• Web

Large-scale (semi-) Automated Analysis
4
Scale

• Large-scale
• (semi)-automated analysis of genome sequence
• Interface with functional genomics
• Medium-scale
• Domain hunting
• Small-scale
• Analysis of individual sequence by power-user

5
(No Transcript)
6
BioinformaticsEnabling Technologies

• Maths
• Algorithm Development
• Computer science
• Ever-increasing list of free software
• Bioinformatics programs
• Operating system LINUX
• Scripting languages glue programs together
• PERL
• Growth of Internet
• Distance is dead!, Distributed Resources
• User-friendliness of Web
• Just Cut and Paste!

7
BioinformaticsChallenges

• DNA Protein Sequences
• Exponential increase
• Genome Sequencing
• Need for annotation
• Molecular stamp collecting?
• Role in drug discovery
• Big business

8
BioinformaticsChallenges

• Interplay between the wet and the dry
• Bioinformatics Predictions
• range from the very general to the very specific
• range from highly speculative to the almost
  certain
• But in the end they are still only predictions
• Need for experimental confirmation

9
Challenges The Data Flood

• 52 bacterial genomes completed and published
• 100,000 genes
• 228 genomes ongoing
• 450,000 more genes when finished

10
The Post-Genomic Iceberg
Discovered Biology
The Undiscovered Genotype Most genes are of
unknown function Undiscovered genomic diversity
Undiscovered Biology
The Undiscovered Phenotype Most bacterial
physiology inapparent in the lab Undiscovered
regulators and regulons
11
Bioinformatics Approaches

• Multiple levels of analysis
• Gene Finding
• Protein function prediction
• Power of homology
• Pitfalls of homology
• Comparative genomics
• Metabolism reconstruction
• Interface with functional genomics

12
What is a Sequence?

• DNA Sequence, double stranded, antiparallel
• Conventionally written 5 to 3
• 5-ATGAGTACCG CTAAATTAGT TAAATCAAAA-3
• 3-TACTCATGGC GATTTAATCA ATTTAGTTTT-5
• RNA sequence, single stranded, U instead of T
• 5-AUGAGUACCG CUAAAUUAGU UAAAUCAAAA-3
• Protein sequence
• conventionally written N-terminal to C-terminal
• 3-letter code Met Ser Thr Ala Lys Leu
• 1-letter code MSTAKLVKSKATN
• Sequences usually written in a monospaced font
  like Courier
• Times Courier
• AGCGGGCGG AGCGGGCGG
• ATCGTTCTG ATCGTTCTG

13
(No Transcript)
14
First get your sequence!

• Most sequencing is now...
• Performed on DNA (rather than RNA or protein)
• Performed using the Sanger didexy method
• Exceptions...
• rRNA is sometimes sequenced directly
• N-terminal and mass spectrometry sequencing of
  proteins
• Template for sequencing can be
• DNA cloned in a plasmid (e.g. pUC19)
• DNA cloned in a single-stranded phage (e.g. M13)
• PCR products

15
First get your sequence!

• Automated Sequencing
• Fluorescent dyes used
• Extract sequence from chromatogram
• Must extract only the error-free region

16
Where to analyse?
17
Assembling a contig
18
Analysis of nucleotide sequence data

• Sequence Composition
• bacteria vary greatly in chromosomal GC
  content each genus has characteristic GC
• useful in
• checking you cloned what you think you have
• identifying foreign DNA in a genome
• Predicted Restriction maps
• helpful in the lab

19
Analysis of nucleotide sequence data

• Search for Other Sequence Features
• Promoters
• Ribosome-binding Sites
• Repeats
• Inverted Repeats (e.g. terminators)
• Consensus Sequences for regulator binding sites

20
Searching for coding regions

• Any given DNA sequence can be translated in 6
  different reading frames, 3 on each strand

21
ORF maps
22
The Problem of Frameshift Errors
Actual sequence
10 20 30 40
50 60 70

ATGAGTACCGCTAAATTAGTTAAATCAAAAGCGACCAATCTGCTTTAT
ACCCGCAACGATGTCTCCGACAGCGAGAAA M S T A K L
V K S K A T N L L Y T R N D V S D
S E K V P L N L N Q K R P I
C F I P A T M S P T A R K E Y
R I S I K S D Q S A L Y P Q R
C L R Q R E K
10 20 30 40
50 60 70

ATGAGTACCGCTAAATTAGTTAAATCAAAAAGCGACCAATCTGCTTTA
TACCCGCAACGATGTCTCCGACAGCGAGAA M S T A K L
V K S K S D Q S A L Y P Q R C L R
Q R E V P L N L N Q K A T N
L L Y T R N D V S D S E K E Y
R I S I K K R P I C F I P A T
M S P T A R K
Frameshifted sequence after single base error
Markov Models (GLIMMER) now commonly used to
predict coding regions
23
Analysis of Protein Sequence Data
24
Analysis of Protein Sequence Data Signal peptides
SignalP uses neural networks
25
Homology

• Similarity that arises because of descent from a
  common ancestor
• The formation of different languages and of
  distinct species, and the proofs that both have
  been developed through a gradual process, are
  curiously parallel We find in distinct languages
  striking homologies due to community of descent,
  and analogies due to a similar process of
  formation Languages, like organic beings, can be
  classed in groups under groups and they can be
  classed either naturally according to descent, or
  artificially by other characters The survival or
  preservation of certain favoured words in the
  struggle for existence is natural selection.
• Charles Darwin, 1871 THE DESCENT OF MAN, Chapter
  3

26
Homology
the cat sat on the mat die Katze sass auf
der Matte
vgeGBant88-2 ITLITCVSVKDNSKRYVVAG vgeGEfae9-1
78 LTLITCDQATKTTGRIIVIA vgeGSpne1-403
MTLITCDPIPTFNKRLLVNF sortase_staur
LTLITCDDYNEKTGVWEKRK
27
Homology
Sequence homology is not just sequence similarity!
Sequences 1, 1A, 1B and 2 are all homologous to
one another Another sequence 2 is similar to
sequences 1, 1A, 1B 2, but not homologous to
them as it does not share a common ancestor with
them Another sequence 1 is neither homologous
nor similar to sequences 1, 1A, 1B 2
28
Types of Homology
29
Sequence Databases

• All sequences when published are deposited in
  Sequence Databases
• Nucleic Acid Sequence Databases
• EMBL, Heidelberg, http//www.embl-heidelberg.de/
• GenBank, in the NCBI, USA, http//www.ncbi.nlm.nih
  .gov/
• Protein Sequence Databases
• GenPept and TREMBL
• Curated database SwissProt, Geneva,
  http//www.expasy.ch/sprot/
• Numerous others, reviewed every year in NAR
• Problem of sequence formats
• Simplest format is FASTA
• gtsequence name
• AATGATGCGTGATGATGATGATGACTGACTGATGATGAT

30
Homology Searches

• The aim of homology searches is to identify
  sequences within these databases that are
  homologous to your sequence.
• This involves comparing your sequence with all
  the database sequences, looking for stretches of
  sequence that appear to be similar, then scoring
  the matches and ranking them. Usually a measure
  of the significance of the match will be given.

31
Homology Searches Translate first!
32
Homology Searches with BLAST

• BLASTN
• Nucleotide query vs nucleotide database
• BLASTP
• protein query vs protein database
• BLASTX
• automatic 6-frame translation of nucleotide query
  vs protein database
• TBLASTN
• protein query vs automatic 6-frame translation of
  nucleotide database
• TBLASTX
• automatic 6-frame translation of nucleotide query
  vs automatic 6-frame translation of nucleotide
  database

33
Typical Blast Output
Sum
Reading High Probability Sequences
producing High-scoring Segment Pairs
Frame Score P(N) N embX69337ECDPS
E.coli dps gene for binding protein 2 834
6.4e-109 1 gbU04242ECU04242 Escherichia
coli core starvation p... 3 828 2.7e-106
1 embX14180ECGLNHPQ Escherichia coli glutamine
permeas... 3 443 2.8e-53
1 gbU18769HDU18769 Haemophilus ducreyi fine
tangled p... 1 150 4.0e-18 2
dbjD01016ANALTI46 Anabaena variabilis lti46
gene. gte... 2 129 4.8e-12 2
gbM84990P26BPO Plasmid pOP2621 ORF1 gene,
5' end... -2 131 6.7e-09
1 gbU16121HPU16121 Helicobacter pylori
neutrophil act... 1 112 1.8e-06
1 gbM32401TRPTYF1 T.pallidum pallidum
antigen TyF1 g... 3 101 5.6e-06
2 embX71436RPNTRB R.phaseoli ntrB gene 1
67 0.76
2 gbL35598DRODGC1A Drosophila melanogaster
receptor g... 1 48 0.97 3
34
Typical Blast Output
gbU18769HDU18769 Haemophilus ducreyi fine
tangled pili major pilin subunit gene Length
780 Plus Strand HSPs Score 150 (68.0 bits),
Expect 4.0e-18, Sum P(2) 4.0e-18 Identities
36/89 (40), Positives 46/89 (51), Frame
1 Query 30 ELLNRQVIQFIDLSLITKQAHWNMRGANFIAVH
EMLDGFRTALIDHLDTMAERAVQLGGV 89 E L
LLI K AHWN G FIAVHEMLD D D AER
LG Sbjct 253 EALQMRLQGLNELALILKHAHWNVVGPQFIA
VHEMLDSQVDEVRDFIDEIAERMATLGVA 432 Query 90
ALGTTQVINSKTPLKSYPLDIHNVQDHLK 118
G YPL QDHLK Sbjct 433
PNGLSGNLVETRQSPEYPLGRATAQDHLK 519
35
Sequence alignments
Dps_trepo .........N MCTDGKKYHS TATSAAVGAS
APGVPDARAI AAICEQLRRH Dps_helico ..........
.......... .......... .......... MKTFEILKHL
Dps_anab .......... .......... .......MPR
INIGLTDEQR QGVINLLNQD MrgA ..........
.......... .......... MKTENAKTNQ TLVENSLNTQ
Dps_haemo MRSKTITFPV LKLTGQSQAL TNDMHKNADH
TVPGLTVATG HLIAEALQMR Dps_ecoli ..........
....MSTAKL VKSKATNLLY TRNDVSDSEK KATVELLNRQ
Dps_strep ........MT SQPHLHQHAA EIQEFGTVTQ
LPIALSHDAR QYSCQRLNRV Dps_trepo VADLGVLYIK
LHNYHWHIYG IEFKQVHELL EEYYVSVTEA FDTIAERLLQ
Dps_helico QADAIVLFMK VHNFHWNVKG TDFFNVHKAT
EEIYEEFADM FDDLAERLVQ Dps_anab LADSYLLLVK
TKKYHWDVVG PQFRSLHQLW EEHYEKLTEN IDAIAERVRT
MrgA LSNWFLLYSK LHRFHWYVKG PHFFTLHEKF
EELYD..... .HAAETWIPS Dps_haemo LQGLNELALI
LKHAHWNVVG PQFIAVHEML DSQVDEVRDF IDEIAERMAT
Dps_ecoli VIQFIDLSLI TKQAHWNMRG ANFIAVHEML
DGFRTALIDH LDTMAERAVQ Dps_strep LADTQFLYAL
YKKCHWGMRG PTAYQLHLLF DKHAQEQLEL VDALAERVQT
ClustalW most commonly used program Note problems
of indels and ragged ends Need for manual
refinement Multiple alignments useful for
identifying active sites and distant homology
36
Into the twilight zoneThe search for distant
homologies
Signal Peptide
A
Proteins consist of domains
B
Signal Peptide
Transitivity of Homology
Coiled coil domain
C
Distant Homology
D
37
Domain Hunting
Homology Search
Add to Alignment
38
PSI-BLAST Position-Specific Iterated BLAST

• combines statistically significant alignments
  produced by BLAST into a position-specific score
  matrix
• searches the database using this matrix
• allows multiple iterations of this process
• runs at approximately the same speed per
  iteration as gapped BLAST
• is much more sensitive to weak but biologically
  relevant sequence similarities

39
Role of Sequence Analysisin the Pre-Genomic Era
Confirm

• Sequence Analysis
• Homology
• Structural Features

Identify Clone Gene
Obtain Sequence
40
Role of Sequence Analysisin the Post-Genomic Era
Obtain Sequence from Genome Project
Formulate Hypothesis

• Sequence Analysis
• Homology
• Structural Features
• Genomic Context

Novel Lab-based Experimental Programme Amplify
Clone Express Gene Create mutant etc.