Bioinformatics

1
Bioinformatics

• CSC 391/691 PHY 392 BICM 715

2
Importance of bioinformatics

• A more global perspective in experimental design
• The ability to capitalize on the emerging
  technology of database-mining--the process by
  which testable hypotheses are generated regarding
  the function or structure of a gene or protein of
  interest by identifying similar sequences in
  better characterized organisms.

3
Amino acids chemical composition or digital
symbols for proteins
http//wbiomed.curtin.edu.au/teach/biochem/tutoria
ls/AAs/AA.html
Link found on the Research Collaboratory for
Structural Biology web site www.rcsb.org/pdb/edu
cation.html
See also Table 2.2 (Mount)
4
Nucleotides chemical composition or digital
symbols for nucleic acids
http//ndbserver.rutgers.edu/NDB/archives/NAintro/

http//www.web-books.com/MoBio/Free/Ch3A.htm
Link found on the Research Collaboratory for
Structural Biology web site www.rcsb.org/pdb/edu
cation.html
See also Table 2.1 (Mount)
5
The Genetic Code how DNA nucleotides encode
protein amino acids
http//www.accessexcellence.org/AB/GG/genetic.html
6
Biologists think its a lot of data, but maybe
its really not
He made fun of biologists for complaining that
the human genome, which takes up about 3
gigabytes, is "a lot of data".  He offered the
comparison of the DVD movie "Evita", which is
about 12 gigabytes, with the genome of Madonna. 
(3 gigabytes).  "The movie contains four times
more information than Madonna's genome.  And
Madonna shares 99 of her DNA with a chimp...And
90 with Craig Venters dog.   More proof that
the genome is not a lot of data  About
90-something percent of genetic information is
common to all humans.  "The unique part of you
will fit on a floppy disk."
Nathan Myhrvold, former Chief Technology Officer
for MicrosoftKeynote Speech at NIH Digital
Biology Meeting 2003
7
Review of Lab 1

• What did you learn about the sites you visited
  SGD, SwissProt, EntrezRefSeq, EntrezNeighbor,
  EntrezProtein, PIR-US
• Can you define the term protein function?
• Does the term gene function have any meaning?
• Questions?

8
Biologists think its a lot of data, but maybe
its really not
He made fun of biologists for complaining that
the human genome, which takes up about 3
gigabytes, is "a lot of data".  He offered the
comparison of the DVD movie "Evita", which is
about 12 gigabytes, with the genome of Madonna. 
(3 gigabytes).  "The movie contains four times
more information than Madonna's genome.  And
Madonna shares 99 of her DNA with a chimp...And
90 with Craig Venters dog.   More proof that
the genome is not a lot of data  About
90-something percent of genetic information is
common to all humans.  "The unique part of you
will fit on a floppy disk."
Nathan Myhrvold, former Chief Technology Officer
for MicrosoftKeynote Speech at NIH Digital
Biology Meeting 2003
9
Biologists think its a lot of data, and maybe it
really is

• The genome is not a static, one-time picture
• Genome changes over timemutations and other
  changes
• Genes expressed to make proteins
• Set of genes that are expressed changes with cell
  type
• Set of genes that are expressed changes over time
  and state

10
Definition of a Biological Database
A biological database is a large, organized body
of persistent data, usually associated with
computerized software designed to update, query,
and retrieve components of the data stored within
the system.
11
Sources of sequence data

• GenBank at the National Center of Biotechnology
  Information, National Library of Medicine,
  Washington, DC (nucleotides and proteins)
  http//www.ncbi.nlm.nih.gov/Entrez
• European Molecular Biology Laboratory (EMBL)
  Outstation at Hixton, England http//www.ebi.ac.uk
  /embl/index.html
• DNA DataBank of Japan (DDBJ) at Mishima, Japan
  http//www.ddbj.nig.ac.jp/
• Protein International Resource (PIR) database at
  the National Biomedical Research Foundation in
  Washington, DC (see Barker et al. 1998)
  http//www-nbrf.georgetown.edu/pirwww/
• The SwissProt protein sequence database at ISREC,
  Swiss Institute for Experimental Cancer Research
  in Epalinges/Lausanne http//www.expasy.ch/cgi-bi
  n/sprot-search-de
• The Sequence Retrieval System (SRS) at the
  European Bioinformatics Institute allows both
  simple and complex concurrent searches of one or
  more sequence databases. The SRS system may also
  be used on a local machine to assist in the
  preparation of local sequence databases.
  http//srs6.ebi.ac.uk

Table 2.5. Mount
12
Sources of protein structure data

• RCSB Protein Data Bank (PDB) www.rcsb.org
• BioMagResBank http//www.bmrb.wisc.edu/
• MMDB http//www.ncbi.nlm.nih.gov/Structure/MMDB/
  mmdb.shtml

13
Review of Lab 2

• What did you learn about the RCSB web page?
• What are your thoughts about the PDB file format?
• Was RasMol easy or hard to use? Is there
  anything you tried to do, but couldnt figure out
  how?
• What is the difference between the two
  glutaredoxin structures (1aaz and 1die)?
• MMDB database of protein structures, ASN.1
  format (http//www.ncbi.nlm.nih.gov/Structure/MMDB
  /mmdb.shtml)
• Other questions?

14
Levels of protein structure

• Primary structure
• Secondary structure
• (Super secondary structure)
• Tertiary structure
• Quaternary structure

15
Databases of protein structure classification

• SCOP Murzin A. G., Brenner S. E., Hubbard T.,
  Chothia C. (1995). J. Mol. Biol. 247, 536-540.
  scop_at_mrc-lmb.cam.ac.uk
• CATH Orengo, C.A., Michie, A.D., Jones, S.,
  Jones, D.T., Swindells, M.B., and Thornton, J.M.
  (1997) Vol 5. No 8. p.1093-1108.
  http//www.biochem.ucl.ac.uk/bsm/cath/
• Dali L. Holm and C. Sander (1996) Science
  273595-602. http//www.bioinfo.biocenter.helsink
  i.fi8080/dali/index.html
• VAST S. H. Bryant and C. Hogue.
  http//www.ncbi.nlm.nih.gov/Structure/VAST/vast.sh
  tml

16
RNA Structure

• Primary structure sequence of GACU nucleotides
• Secondary structure stem-loop structures
• Tertiary structure
• http//www.rnabase.org/

17
DNA structure

• Primary structure sequence of GACT nucleotides
• Secondary structure double helix
• Higher levels of structure nucleosome
  chromatin chromosome

18
An example of pairwise alignment

• ./wwwtmp/lalign/.17728.1.seq Glutaredoxin, T4,
  1AAZ.pdb - 87 aa
• (B) ./wwwtmp/lalign/.17728.2.seq Unknown protein
  - 93 aa
• using matrix file BL50, gap penalties -14/-4
• 27.0 identity in 89 aa overlap score 101
  E(10,000) 0.0014

10 20 30 40
50 Glutar KVYGYDSNIHKCVYCDNAKRLLTVK
KQPFEFINIMPEKGV---FDDEKIAELLTKLGR ..
.. . .. .. . . .
.. . Unknow EIYGIPEDVAKCSGCISAIRLCFEKGYDYEIIPVLKK
ANNQLGFDYILEKFDECKARANM 10 20
30 40 50 60
60 70 80 Glutar
DTQIGLTMPQVFAPDGSHIGGFDQLREYF ..
..... .... ... .Unknow QTR-PTSFPRIFV-DGQYI
GSLKQFKDLY 70 80 90
19
Pairwise Sequence Alignment

• The alignment of two sequences (either protein or
  nucleic acid) based on some algorithm
• What is the right answer?
• Align (pairwise) the following words
  instruction, insurrection, incision
• There is NO unique, precise, and universally
  applicable method of pairwise alignment

20
An example of pairwise alignment

• ./wwwtmp/lalign/.17728.1.seq Glutaredoxin, T4,
  1AAZ.pdb - 87 aa
• (B) ./wwwtmp/lalign/.17728.2.seq Unknown protein
  - 93 aa
• using matrix file BL50, gap penalties -14/-4
• 27.0 identity in 89 aa overlap score 101
  E(10,000) 0.0014

10 20 30 40
50 Glutar KVYGYDSNIHKCVYCDNAKRLLTVK
KQPFEFINIMPEKGV---FDDEKIAELLTKLGR ..
.. . .. .. . . .
.. . Unknow EIYGIPEDVAKCSGCISAIRLCFEKGYDYEIIPVLKK
ANNQLGFDYILEKFDECKARANM 10 20
30 40 50 60
60 70 80 Glutar
DTQIGLTMPQVFAPDGSHIGGFDQLREYF ..
..... .... ... .Unknow QTR-PTSFPRIFV-DGQYI
GSLKQFKDLY 70 80 90
21
An example of pairwise alignment

• ./wwwtmp/lalign/.17731.1.seq unknown protein,
  Arabidopsis - 201 aa
• /wwwtmp/lalign/.17731.2.seq Heamophilus Influenza
  Hybrid-Prx5, 1nm3 - 241 aa using matrix file
  BL50, gap penalties -14/-4
• 36.4 identity in 140 aa overlap score 288
  E(10,000) 3.5e-19

70 80 90 100 110
120unknow KFSTTPLSDIFKGKKVVIFGLPGAYTGVCSQQHVPSYKS
HIDKFKAKGIDSVICVSVNDP . .
.. . .... .... . ....
Heamop KWVDVTTSELFDNKTVIVFSLPGAFTPTCSSSHLPR
YNELAPVFKKYGVDDILVVSVNDT 30
40 50 60 70 80
130 140 150 160
170 180unknow FAINGWAEKLGAKDAIEFYGDFDGKFHKS
LGLDKDLSAALLGPRSERWSAYVEDGKVKAV . .
... .. .. . . .. .
Heamop FVXNAWKEDEKSEN-ISFIPDGNGEFTEGXGXLVGKEDLGFGK
RSWRYSXLVKNGVVEKX 90 100
110 120 130
190 unknow NVE-EAPSD-FKVTGAEVIL
. . . . .. Heamop FIEPNEPGDPFKVSDADTXL
140 150
22
Global vs Local Alignment
Figure 3.1, Mount
23
Pairwise Sequence Alignment Websites
Bayes block aligner http//www.wadsworth.org/resres/bioinfo Zhu et al. (1998)
BCM Search Launcher Pairwise sequence alignment http//searchlauncher.bcm.tmc.edu/seq-search/alignment.html  
SIMLocal similarity program for finding alternative alignments http//www.expasy.ch/tools/sim.html Huang et al. (1990) Huang and Miller (1991) Pearson and Miller (1992)
Global alignment programs (GAP, NAP) http//genome.cs.mtu.edu/align/align.html Huang (1994)
FASTA program suite http//fasta.bioch.virginia.edu/fasta/fasta_list.html Pearson and Miller (1992) Pearson (1996)
BLAST 2 sequence alignment (BLASTN, BLASTP) http//www.ncbi.nlm.nih.gov/gorf/bl2.html Altschul et al. (1990)
LALIGN http//www.ch.embnet.org/software/LALIGN_form.html Huang and Miller, published in Adv. Appl. Math. (1991) 12337-357
Likelihood-weighted sequence alignment (lwa) http//stateslab.bioinformatics.med.umich.edu/service/lwa.html  
Table 3.1, Mount
24
What is multiple sequence alignment?

• Multiple sequence alignment is the alignment of
  more than two nucleotide or protein sequences
• Compare pairwise sequence alignment multiple
  sequence alignment

25
Issues with multiple sequence alignment

• Try creating a multiple sequence alignment of the
  three words
• Insurrection
• Incision
• Instruction

26
Issues with multiple sequence alignment

• Whats the right answer?
• Computational complexity
• What is reasonable method for obtaining
  cumulative score?
• Placement and scoring of gaps

in cisioninsurrec tioninstr uc tion
in cisioninsurrectionins truction
in cisioninsurrectioninstr uction
inci sioninsurrectionins truction
27
Pairwise sequence alignment LALIGN of OVCA2 and
DYR_SCHPO (global)
./wwwt MAAQRPLRVLCLAGFRQSERGFREKTGALRKALRGRAELVCLS
GPHPVPDPPGPEGARSD . .. .
. ... . . . . . dihydr
MSKPLKVLCLHGWIQSGPVFSKKMGSVQKYLSKYAELHFPTGPVVADEE
ADPNDEEEK 10 20 30
40 50 70
80 90 100 110
120./wwwt FGSCPPEEQPRGWWFSEQEADVFSALEEPAVCRGLEESL
GMVAQALNRLGPFDGLLGFSQ . .
.. . . . . . ... .dihydr
KRLAALGGEQNGGKFGWFEVEDFKN-----TYGSWDESLECINQYMQEKG
PFDGLIGFSQ 60 70 80
90 100 110 130
140 150 160 170
./wwwt GAALAALVCALGQAGDPRFPL---PRFILLVSSFCPRGIGF
KESILQRPLSLPSLHVF ..... . .
...... . . . . . . dihydr
GAGIGAMLAQMLQPGQPPNPYVQHPPFKFVVFVGGFRAEKPEF-DHFYNP
KLTTPSLHIA 120 130 140
150 160 170 180
190 200 210
./wwwt GDTDKVIPSQESVQLASQFPGAITLTHSGGHFIPA-------
------AAP--------- . .. . . . .
. .. . dihydr
GTSDTLVPLARSKQLVERCENAHVLLHPGQHIVPQQAVYKTGIRDFMFSA
PTKEPTKHPR
19.2 sequence identity score -413
28
Multiple sequence alignment
29
What is multiple sequence alignment used for?

• Consensus sequences which residues can be used
  to identify other members of the family?
• Gene and protein families which residues are
  functionally important functional families
• Relationships and phylogenies contains
  evolutionary history of sequences
• Data underlying some protein structure prediction
  algorithms
• Genome sequencing sequence random, overlapping
  fragments automation of assembly (in this case,
  there is a RIGHT answer)

30
Consensus sequences and important functional
residues
Baxter, et al, Mol Cell Prot 2003
31
Relationships and phylogenies

• Serine-threonine protein phosphatases
• Same biochemical function
• Clustering clearly shows PP1, PP2a and PP2B
  families
• What is different about these families?

Fetrow, Siew, Skolnick, FASEB J, 1999
32
Possible redox site in PP1 family
Only a clustering, not a true phylogenetic tree
33
Methods to solve computational complexity

• Progressive global alignment
• Iterative methods
• Alignments based on locally conserved patterns
• Statistical methods and probabilistic models

34
Multiple Sequence Alignment Global
CLUSTALW or CLUSTALX (latter has graphical interface) FTP to ftp.ebi.ac.uk/pub/softwarea,d Thompson et al. (1994a, 1997) Higgins et al. (1996)
MSA http//www.psc.edu/bhttp//www.ibc.wustl.edu/ibc/msa.htmlcFTP to fastlink.nih.gov/pub/msa Lipman et al. (1989)Gupta et al. (1995)
PRALINE http//mathbio.nimr.mrc.ac.uk/jhering/praline/ Heringa (1999)
Table 4.1, Mount
35
Multiple Sequence Alignment Interative
 
DIALIGN segment alignment http//www.gsf.de/biodv/dialign.html   Morgenstern et al. (1996)
MultAlin http//protein.toulouse.inra.fr/multalin.html   Corpet (1988)
Parallel PRRN progressive global alignment http//prrn.ims.u-tokyo.ac.jp/   Gotoh (1996)
SAGA genetic algorithm http//igs-server.cnrs-mrs.fr/cnotred/Projects_home_page/saga_home_page.html   Notredame and Higgins (1996)
Table 4.1, Mount
36
Multiple Sequence Alignment Local
Aligned Segment Statistical Evaluation Tool (Asset) FTP to ncbi.nlm.nih.gov/pub/neuwald/asset Neuwald and Green (1994)
BLOCKS Web site http//blocks.fhcrc.org/blocks/ Henikoff and Henikoff (1991, 1992)
eMOTIF Web server http//dna.Stanford.EDU/emotif/ Nevill-Manning et al. (1998)
GIBBS, the Gibbs sampler statistical method FTP to ncbi.nlm.nih.gov/pub/neuwald/gibbs9_95/ Lawrence et al. (1993) Liu et al. (1995) Neuwald et al. (1995)
HMMER hidden Markov model software http//hmmer.wustl.edu/ Eddy (1998)
MACAW, a workbench for multiple alignment construction and analysis FTP to ncbi.nlm.nih.gov/pub/macaw/ Schuler et al. (1991)
MEME Web site, expectation maximization method http//meme.sdsc.edu/meme/website/ Bailey and Elkan (1995) Grundy et al. (1996, 1997) Bailey and Gribskov (1998)
Profile analysis at UCSDa,e http//www.sdsc.edu/projects/profile/ Gribskov and Veretnik (1996)
SAM hidden Markov model Web site http//www.cse.ucsc.edu/research/compbio/sam.html Krogh et al. (1994) Hughey and Krogh (1996)
Table 4.1, Mount
37
Methods to solve computational complexity

• Progressive global alignment
• Start with most related sequences
• Problem is that these errors in initial
  alignments are propagated
• Iterative methods
• Iterative alignment of subgroup of sequences to
  find best then align subgroups
• Alignments based on locally conserved patterns
• Block analysis
• Statistical methods and probabilistic models
• Expectation maximum Gibbs sampler Hidden Markov
  Models

38
Profile Methods

• Perform a global multiple sequence alignment on a
  group of sequences
• Extract more highly conserved regions
• Profile scoring matrix for these highly
  conserved regions
• Used to search unknown sequences for membership
  in the family

Figures 4.11 (p. 162) and 4.12 (p. 166-167)
39
Limitations of such profiles

• Limited by sequences in original msa
• Sequence bias (too many of one type of sequence)
• Sequences in msa not representative of entire
  family

40
Blocks

• Blocks are conserved regions of msa (like
  profiles) but no gaps allowed
• Servers for producing Blocks
• Blocks server
• eMotif server
• Block libraries for database searching
• Blocks (Henikoff and Henikoff)
• Prosite (Bairoch)
• Prints (Attwood)

41
Blocks that might be extracted from an msa
Baxter, et al, Mol Cell Prot 2003
42
Blocks that might be extracted from an msa
Baxter, et al, Mol Cell Prot 2003
43
Database searching

• Identify a new sequence by experimental methods
  what is it?
• Search databases to find similar sequences
• If enough similarity, can say that function of
  new sequence is same as known sequence function
  annotation transfer
• What is enough similarity?
• What is function?

Chapter 7, Mount
44
Relationships between family members

• Sequence relationships between family members
• Not all members of family have significant
  sequence similarity to all others
• Can be represented by nodes and edges of a graph

Z
F
E
A
D
C
B
45
Beware of issues with function annotation transfer

• Multiple domains
• High sequence identity, but functional residues
  not conserved
• Sequence repeats (low complexity regions)

New
Function B Function A
Function A
H
S
D
Known serine hydrolase
New sequence
S
D
L
46
Methods for database searching

• Sequence similarity with query sequence FASTA,
  BLAST (Fig 7.5, p. 305)
• Profile search ProfileSearch
• Position-specific scoring matrix MAST
• Iterative alignment (combination of sequence
  searching and profile search) PSI-BLAST
• Patterns Prosite, Blocks, Prints, CDD/Impala

Table 7.1, Mount
47
The problem with speed

• Dynamic programming
• Guaranteed to find optimal answer
• Too slow (number of searches performed and number
  of sequences in databases that are searched)
  Smith-Waterman dynamic programming algorithm 50X
  slower than BLAST or FASTA
• faster hardware has made this problem feasible
• Heuristic methods
• FASTA short, common patterns in query and
  database searches
• BLAST similar, but searched for more rare and
  significant patterns

48
Searches on DNA vs Protein Sequences

• 20-letter alphabet vs 4-letter alphabet
• Fivefold larger variety of sequence characters in
  proteins easier to detect patterns
• Searches with DNA sequences produce fewer
  significant matches
• What if you dont know reading frame?
• Sometimes must do nucleic acid searches
  (searching for similarities in non-coding regions)

49
Sensitivity vs selectivity

• Sensitivity methods ability to find most
  members of the protein family
• Selectivity methods ability to distinguish
  true members from non-members
• Want a method to have high sensitivity (get all
  true positives) and high selectivity (not get
  false positives)
• Can be a difficult test with biological data
  sets not all true positives are known

50
Scoring matrices commonly used

• PAM250 point accepted mutation Dayhoff, M.,
  Schwartz, R. M., and Orcutt, B. C., Atlas of
  Protein Sequence and Structure (1978) 5(3)345
• BLOSUM62 blocks amino acid substitution
  matrices Henikoff and Henikoff, Amino acid
  substitution matrices from protein blocks. (1992)
  Proc. Natl. Acad. Sci. USA 8910915-10919.

51
PAM250

• Calculated for families of related proteins (gt85
  identity)
• 1 PAM is the amount of evolutionary change that
  yields, on average, one substitution in 100 amino
  acid residues
• A positive score signifies a common replacement
  whereas a negative score signifies an unlikely
  replacement
• PAM250 matrix assumes/is optimized for sequences
  separated by 250 PAM, i.e. 250 substitutions in
  100 amino acids (longer evolutionary time)

52
BLOSUM62

• BLOSUM matrices are based on local alignments
  (blocks or conserved amino acid patterns)
• BLOSUM 62 is a matrix calculated from comparisons
  of sequences with no less than 62 divergence
• All BLOSUM matrices are based on observed
  alignments they are not extrapolated from
  comparisons of closely related proteins
• BLOSUM 62 is the default matrix in BLAST 2.0

53
Comparison of PAM250 and BLOSUM62
BLOSUM80 PAM1
BLOSUM62 PAM120
BLOSUM45 PAM250
Less divergent
More divergent
The relationship between BLOSUM and PAM
substitution matrices. BLOSUM matrices with
higher numbers and PAM matrices with low numbers
are both designed for comparisons of closely
related sequences. BLOSUM matrices with low
numbers and PAM matrices with high numbers are
designed for comparisons of distantly related
proteins. If distant relatives of the query
sequence are specifically being sought, the
matrix can be tailored to that type of search.
54
Scoring matrices commonly used

• PAM250
• Represents a period of time during which only
  about 20 of amino acids will remain unchanged
• Shown to be appropriate for searching for
  sequences of 17-27 identity
• BLOSUM62
• Matrix calculated from comparisons of sequences
  with no less than 62 divergence
• Though it is tailored for comparisons of
  moderately distant proteins, it performs well in
  detecting closer relationships
• BLOSUM50
• Shown to be better for FASTA searches

55
Methods for database sequence searching

• Sequence similarity with query sequence FASTA,
  BLAST
• Profile search ProfileSearch
• Position-specific scoring matrix MAST
• Iterative alignment (combination of sequence
  searching and profile search) PSI-BLAST
• Patterns Prosite, PFAM, CDD/Impala

56
Review of protein structure

• Primary structure sequence of amino acids
• Secondary structure local segments of protein
  structure
• Tertiary structure three-dimensional structure
  of a single protein chain
• Quaternary structure packing of 2 or more
  protein chains

57
Classification of protein tertiary structure

• All alpha proteins
• All beta proteins
• Alphabeta proteins
• Alpha/beta proteins
• Irregular proteins

Classify these proteins T-cell protein CD8
(1cd8), myoglobin, triose phosphate isomerase,
G-specific endonuclease (1rnb)
58
Representations of protein structures

• All atom
• CPK models
• Cartoons (ribbons, etc)
• Topology diagrams

59
Protein structure databases

• RCSB (PDB) http//www.rcsb.org/pdb
• General repository for all protein coordinate
  files
• MMDB http//www.ncbi.nlm.nih.gov/Structure
• NCBI structure database structures from pdb
• Links to sequence and genome databases
• BioMagResBank http//www.bmrb.wisc.edu/
• General repository for NMR structure data

60
Alignment of protein structure

• Superposition of protein 3D structures
• Used in searching for structural similarity and
  grouping proteins into fold families
• Structural similarity is common and does not
  necessarily indicate an evolutionary relationship
  (different from sequence similarity)

61
Structure Alignment A difficult problem

• Alignment in atom positions in 3D space
• Pieces of proteins may align
• What is significant and what is not? (Is
  alignment of two helices significant?)
• Alignment of topology or secondary structure
  packing give different answers

Easy example (Eidhammer and Jonassen)
More difficult examples http//www.sbg.bio.ic.ac.
uk/people/rob/sf/sf.html
62
Structure alignment used to classify (group)
protein structures

• SCOP (Structural Classification Of Proteins
  http//scop.mrc-lmb.cam.ac.uk/scop/)
• Class (all alpha, all beta, alphabeta,
  alpha/beta), family, superfamily, fold
• Reflects structural and evolutionary
  relationships
• Mostly done by hand (expert analysis)
• CATH (classification by class, architecture,
  topology and homology http//www.biochem.ucl.ac.u
  k/bsm/cath)
• Class (all alpha, all beta, alpha/beta),
  architecture, fold, superfamily, family
• Uses SSAP structure alignment program
• FSSP (fold classification based on
  structure-structure protein alignment
  http//www.bioinfo.biocenter.helsinki.fi8080/dali
  /index.html.)
• Based on pairwise alignment of all non-redundant
  proteins in PDB
• Divides proteins into structures and domains
  represents unique configuration of secondary
  structure elements
• Uses Dali structure alignment program
• MMDB (molecular modeling database
  http//www.ncbi.nlm.nih.gov/entrez/query.fcgi?dbS
  tructure)
• Proteins classified into structurally related
  groups by VAST, based on arrangements of
  secondary structures
• Groupings of all PDB structures
• SARF (spatial arrangement of backbone fragments
  http//123d.ncifcrf.gov/)

63
Web sites for structure alignment

• VAST http//www.ncbi.nlm.nih.gov/Structure/VAST/
  vast.shtml
• NCBI structure comparison
• Comparison of orientations of secondary
  structures (vector representation of secondary
  structures)
• Approach from graph theory
• Dali http//www.ebi.ac.uk/dali/
• FSSP structure comparison
• Protein represented as distance matrix between
  alpha carbons
• Monte Carlo simulation to do random search for
  sub-distance-matrices
• SSAP http//www.biochem.ucl.ac.uk/cgi-bin/cath/G
  etSsapRasmol.pl
• CATH structure comparison
• Set structure environment for each residue, then
  align residue by residue using double dynamic
  programming
• Structure environment can use beta carbon vectors
  or phi/psi backbone dihedral angles
• Others Lots, such as Structal (Gerstein and
  Levitt) Minarea (Falicov and Cohen) Lock (Singh
  and Brutlag)

64
Protein Structure Prediction

• Goal is to understand the relationship between
  the primary amino acid sequence and the structure
  of the protein
• Relationship between sequence and structure is
  not simple and is not understood
• Protein folding problem remains unsolved

65
Protein Structure Prediction

• Secondary structure prediction unsolved?
• Tertiary structure prediction unsolved problem
  (CASP competition)
• Quaternary structure prediction unsolved
  problem
• Docking of two subunits

66
Secondary structure prediction

• Prediction of three classes of secondary
  structure helix, strand, coil
• Solved problem? 70-80 correct predictions
• Methods (web sites) can give very different
  answers
• Prediction of non-regular secondary structure
  (loops and turns) not as successful

67
Secondary structure prediction

• Method development
• Frequencies on types of residues found in each
  secondary structures
• Frequencies calculated from database of known
  structures (training set)
• Method evaluation
• Test method on proteins whose structures are
  known (testing set)
• Training and testing sets must not be the same

68
Secondary structure prediction methods and
references
Single residue statistics Explicit rules Nearest neighbors Neural networks Hidden markov models
1st generation Chou/Fasman (74) GOR I Lim (74)
2nd generation GOR III (87) Predator (96) Levin (86) Nishikawa and Ooi (86) Yi and Lander (93) Qian and Sejnowski (88) Holley and Karplus (89) Yi and Lander (93) Asai/Handa (93)
3rd generation GOR IV DSC (Prof) (96) NNssp (95) NNssp (95) PHD (93) Jnet (99) PsiPred (99) PASSML (98)
See Table 9.7, Mount, for list of servers
69
GOR IV secondary structure prediction

• Three state prediction helix, strand, loop
• Statistics of pair frequencies observed within a
  window of 17 amino acid residues
• Based on information theorysound statistical
  basis and no ad hoc rules
• Mean accuracy of 64.4 for a three state
  prediction (Q3)

Garnier, Gibrat, Robson http//abs.cit.nih.gov/in
dex.html
70
PHD secondary structure prediction

• Three state prediction helix, strand, loop
• Predicts secondary structure from multiple
  sequence alignments
• Three consecutive neural networks (feed forward)
• Raw 3-state prediction for each position, based
  on alignment composition in 13 residue window
• Filter 3-state probabilities based on
  probabilities of flanking positions in 17-residue
  window
• Jury network using several raw/filter
  combinations trained separately
• Expected average accuracy gt 72 for three state
  prediction (Q3)

Rost and Sander http//www.predictprotein.org
71
Method evaluation how good is good?

• Testing of prediction methods involves
• Applying the method to a set of proteins whose
  secondary structures are known experimentally and
  comparing prediction results to known results
• Calculating measures of how good the performance
  is
• Q1 (h, s, or c)
• (number of residues correctly predicted in one
  state/number of residues in that state) 100
• Q3 (h, s, and c)
• (number of residues correctly predicted in each
  of 3 states/number of all residues) 100
• Matthews correlation coefficient (Cs)
• (TpTn - FpFn) / sqrt(TpFp)(TnFn)(TpFn)(TnFp)
  

Num ....,....1....,....2....,....3....,....4....
,....5....,....6 Res MSTKQHSAKDELLYLNKAVVFGSGAFG
TALAMVLSKKCREVCVWHMNEEEVRLVNEKREN Actu
HHHHHHHHHHHH EE HHHHHHHHHHHHHHH EE
HHHHHHHHHHHHHH Pred HHHHHH EEEEE
HHHHHHHHHHHH EEEEEE HHHHHHHH
Pred HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
HHHHHHHHHHHHHHHH
72
Method evaluation how good is good?

• Matthews correlation coefficient (Cs)
• (TpTn - FpFn) / sqrt(TpFp)(TnFn)(TpFn)(TnFp)
  
• Where Tp, true positive predictions (method
  predicts helix, and residue is in a helix) Tn,
  true negative prediction (method predicts not
  helix, and residue is not in a helix) Fp,
  false positive prediction (method predicts helix,
  but residue is not in a helix) Fn, false
  negative prediction (method predicts not helix,
  but residue is in a helix)

Num ....,....1....,....2....,....3....,....4....
,....5....,....6 Res MSTKQHSAKDELLYLNKAVVFGSGAFG
TALAMVLSKKCREVCVWHMNEEEVRLVNEKREN Actu
HHHHHHHHHHHH EE HHHHHHHHHHHHHHH EE
HHHHHHHHHHHHHH Pred HHHHHH EEEEE
HHHHHHHHHHHH EEEEEE HHHHHHHH
Q1 (helix)(4128)/(121514)10058 Q3(4128
22)/6010047 Tp412824 Tn9817 Fp2
Fn8125117 Ch(2417)-(217)/sqrt(242)(17
17)(2417)(172)
73
Tertiary Structure Prediction

• Homology modeling identifiable sequence
  similarity
• Fold recognition (threading Table 9.8 for
  server list)
• Ab initio methods

74
Homology modeling

• Sequence alignment
• Side chain modeling
• Modeling insertions and deletions
• Optimizing the model
• Model evaluation
• Repeat?

75
Fold Recognition (threading)

• Template identification/sequence
  alignment/alignment optimization
• Side chain modeling
• Modeling insertions and deletions
• Optimizing the model
• Model evaluation
• Repeat?

76
Ab initio methods folding from scratch

• Start with unfolded protein or random
  conformation
• Use atomic-level forces, solve energetic
  equations
• Identify most stable conformation (lowest free
  energy)
• Computational demands high for protein of 100
  amino acids
• Assume constant bond lengths and angles
• Allow 2/3 backbone torsion angles per amino acid
  to rotate
• Do not allow side chain torsion angles to move
• Assuming 10 allowed conformations per residue,
  must explore 10100 conformations
• Calculation of 10100 energies (one for each
  conformation) is not possible

77
Ab initio methods simplifications

• Lattice models to simplify the conformational
  search space
• Monte Carlo statistical sampling of
  conformational space
• Stepwise processes
• Predict regular secondary structures
• Pack secondary structures to form tertiary
  structures
• Others

78
Review of Definitions

• Cell fundamental working unit of biology
• DNA encodes all information to create cells and
  allow them to function
• Linear arrangement of bases (AGTC)
• Genome organisms complete set of DNA
• Chromosome physically distinct molecules of DNA
• Genomes can be composed of 1, 2 or more
  chromosomes
• Gene basic physical and functional unit of
  heredity
• Linear arrangement of bases along the chromosome
• Contain instructions for encoding protein
• (Remember genetic code?)

79
Genomes and proteomes

• Genome Sum of all genes and intergenic DNA
  sequences in a cell
• the smallest known genome for a free-living
  organism (a bacterium) contains about 600,000 DNA
  base pairs
• human and mouse genomes have about 3 billion
• relatively unchanging from cell to cell
• Proteome The entire set of proteins encoded in
  the genome of an organism and produced by that
  organism
• Constellation of proteins in cells is highly
  dynamic

80
The Human Genome

• 24 chromosomes
• Chromosomes range is size from 50 million to 250
  million base pairs
• Total size of the human genome is over 3 billion
  base pairs (3.1647 billion)
• 99.9 of all bases are the same in all people
• Genes comprise only 2 of the total genome
• Human genome is estimated to contain 30,000 to
  40,000 genes
• Average gene size is about 3000 bases
• Largest identified so far is 2.4 million bases
  (dystrophin)
• Functions for less than 50 of genes and gene
  products are known
• Remainder of genome is non-coding regions
• Chromosomal structural integrity
• Repetitive sequences
• Regulation of protein production
• Other functions that we dont know about

81
Human Genome Sequencing Project Goals

• Determine the sequences of the 3 billion chemical
  base pairs that make up human DNA
• Identify all the approximately 30,000 genes in
  human DNA
• Store this information in databases
• Improve tools for data analysis
• Transfer related technologies to the private
  sector
• Address the ethical, legal, and social issues
  that may arise from the project

Human Genome Project (DOE) http//www.ornl.gov/sc
i/techresources/Human_Genome/home.shtml NIH http
//www.ncbi.nlm.nih.gov/genome/guide/human/
82
Other sequencing projects

• Over 200 genomes sequenced
• Range of archeae, bacteria, eukaryotic genomes
• Organisms that have been well-studied in the
  laboratory
• Organisms that are pathogenic to humans
• Organisms of special scientific or technical
  interest

NCBI list of sequenced genomes (NIH) http//www.n
cbi.nlm.nih.gov/entrez/query.fcgi?dbGenome
83
Prokaryotes and eukaryotes

• Prokaryotes (bacteria and archaea)
• No true nucleus
• DNA generally circular (one chromosome)
• Eukaryotes
• True nucleus contains (most) DNA
• DNA linear and arranged in chromosomes

Phylogenetic analysis of small subunit ribosomal
RNAs, C. Woese, 1987
84
Anatomy of a prokaryotic genome

• DNA compact and circular
• ORFs (open reading frames) with start and stop
  codons
• No introns

85
Anatomy of a eukaryotic genome

• Linear DNA chromosomes
• Centromeres
• Telomeres
• Tandem repeats
• Transposable elements
• Introns
• Pseudogenes

Example of chromosome maps http//www.ncbi.nlm.ni
h.gov/genome/guide/human/
86
DNA sequencing
A G C T

• Separate strands of DNA
• Anneal primer to one strand
• Replicate using fluorescently labeled ddNTPs (as
  opposed to normal dNTPs)
• Separate fragments by size
• Image gel for fluorescent labels

See also, electropherogram, Fig2.2, Mount
87
Methods of genome sequencing

• Mapping method
• Fragment chromosome
• Identify markers and order them
• Arrange fragments, then sequence
• Shotgun method
• Fragment chromosome
• Sequence fragments, then arrange
• cDNA sequencing (ESTs)
• Isolate mRNA (expressed in cell)
• Reverse transcribe mRNA to create cDNA
• Sequence cDNA

88
Maps

• Gene map
• Chromosome map
• Sequence map
• Maps important for obtaining sequence information
  (mapping method)
• Restriction map
• Contig (contiguous clone) map

NCBI map viewer http//www.ncbi.nlm.nih.gov/mapvi
ew/
89
Prediction of genes

• Method
• Difference between prokaryotes and eukaryotes
• Tests for validation of predictions

90
Genome Analysis

• General approach (p. 492)
• Comparative genomics
• Self-comparison reveals gene families and
  duplication
• Between-genome-comparison reveals orthologs, gene
  families and domains
• Gene ordering on chromosomes
• Phylogenetic analysis
• Genetic diversity