Transcript Three main topics for this Intro lecture

Simple Rearrangements
Reversals
1
2
3
9
8
4
7
1, 2, 3, 4, 5, 6, 7, 8, 9, 10
•
Blocks represent conserved genes.
6
5
10
1
2
Reversals
3
9
8
4
7
1, 2, 3, -8, -7, -6, -5, -4, 9, 10


10
6
5
Blocks represent conserved genes.
In the course of evolution or in a clinical context, blocks 1,…,10
could be misread as 1, 2, 3, -8, -7, -6, -5, -4, 9, 10.
Types of Rearrangements
Reversal
1 2 3 4 5 6
1 2 -5 -4 -3 6
Translocation
1 2 3
45 6
1 26
4 53
Fusion
1 2 3 4
5 6
1 2 3 4 5 6
Fission
Sorting by reversals: 5 steps
Step
Step
Step
Step
Step
Step
0: p 2 -4
1:
2 3
2:
2 3
3:
2 3
4:
-8 -7
5: g 1 2
-3
4
4
4
-6
3
5
5
5
5
-5
4
-8
-8
6
6
-4
5
-7
-7
7
7
-3
6
-6
-6
8
8
-2
7
1
1
1
-1
-1
8
Sorting by reversals: 4 steps
Step
Step
Step
Step
Step
0: p 2 -4 -3
1:
2 3 4
2:
-5 -4 -3
3:
-5 -4 -3
4: g 1 2 3
5
5
-2
-2
4
-8
-8
-8
-1
5
-7
-7
-7
6
6
-6
-6
-6
7
7
1
1
1
8
8
Sorting by reversals: 4 steps
Step
Step
Step
Step
Step
0: p 2 -4 -3
1:
2 3 4
2:
-5 -4 -3
3:
-5 -4 -3
4: g 1 2 3
5
5
-2
-2
4
-8
-8
-8
-1
5
-7
-7
-7
6
6
-6
-6
-6
7
7
1
1
1
8
8
What is the reversal distance for this
permutation? Can it be sorted in 3 steps?
From Signed to Unsigned Permutation (Continued)
• Construct the breakpoint graph as usual
• Notice the alternating cycles in the graph between every other vertex
pair
• Since these cycles came from the same signed vertex, we will not be
performing any reversal on both pairs at the same time; therefore, these
cycles can be removed from the graph
0
5
6 10 9 15 16 12 11 7 8 14 13 17 18 3
4
1 2 19 20 22 21 23
Reversal Distance with Hurdles
• Hurdles are obstacles in the genome rearrangement problem
• They cause a higher number of required reversals for a permutation
to transform into the identity permutation
• Let h(π) be the number of hurdles in permutation π
• Taking into account of hurdles, the following formula gives a
tighter bound on reversal distance:
d(π) ≥ n+1 – c(π) + h(π)
Median Problem
Goal: find M so that DAM+DBM+DCM is minimized
NP hard for most metric distances
Genome Enumeration for
Multichromosome Genomes
.
.
.
Genome Enumeration
For genomes on gene {1,2,3}
2
.
.
.
-3
1
-1
$
23
.
.
.
2
.
.
.
3
$
‹ 1, 2, 3 ›
-3
$
‹ 1, 2, -3 ›
3
$
‹ 1, 2 › ‹ 3 ›
-3
$
‹ 1, 2 › ‹ -3 ›
...
...
-2-3
...
3
...
-3
...
Rearrangement Phylogeny
Compute A Given Tree (Start)
Compute A Given Tree (First Median)
Compute A Given Tree (Second Median)
Compute A Given Tree (Third Median)
Compute A Given Tree (After 1st Iteration)
Binary Encoding
MLBE Sequences
Experimental Results (Equal Content)
80% inversion, 20% transposition
An Example—New Genomes
1 2
1 -4
…
3
5
1 3 5 7 9
1 5 9 -7 3
…
4
2
5 6
8 10
7 8 9 10
9 -7 -6 3
Jackknifing Rate
Support Value Threshold - FP
Up to 90% FP can be identified with 85% as the
threshold
Jackknife Properties
• Jackknifing is necessary and useful for gene
order phylogeny, and a large number of
errors can be identified
• 40% jackknifing rate is reasonable
• 85% is a conservative threshold, 75% can
also be used
• Low support branches should be examined
in detail
Protein
In-silico Biochemistry
• Online servers exist to determine many
properties of your protein sequences
• Molecular weight
• Extinction coefficients
• Half-life
• It is also possible to simulate protease digestion
• All these analysis programs are available on
• www.expasy.ch
Analyzing Local Properties
• Many local properties are important for the function of
your protein
• Hydrophobic regions are potential transmembrane domains
• Coiled-coiled regions are potential protein-interaction
domains
• Hydrophilic stretches are potential loops
• You can discover these regions
• Using sliding-widow techniques (easy)
• Using prediction methods such as hidden Markov Models
(more sophisticated)
Sliding-window Techniques
• Ideal for identifying strong
signals
• Very simple methods
• Few artifacts
• Not very sensitive
• Use ProtScale on
www.expasy.org
• Make the window the same
size as the feature you’re
looking for
www.expasy.org/cgi-bin/protscale.pl
www.expasy.org/cgi-bin/protscale.pl
www.expasy.org/cgi-bin/protscale.pl
www.expasy.org/cgi-bin/protscale.pl
Hphob. / Eisenberg
Transmembrane Domains
• Discovering a transmembrane
domain tells you a lot about your
protein
• Many important receptors have 7
transmembrane domains
• Transmembrane segments can be
found using ProtScale
• The most accurate predictions
come from using TMHMM
Using TMHMM
• TMHMM is the best method for predicting transmembrane
domains
• TMHMM uses an HMM
• Its principle is very different from that of ProtScale
• TMHMM output is a prediction
TMHMM vs. ProtScale
>sp|P78588|FREL_CANAX Probable ferric reductase transmembrane component OS=Candida albicans
GN=CFL1 PE=3 SV=1
MTESKFHAKYDKIQAEFKTNGTEYAKMTTKSSSGSKTSTSASKSSKSTGSSNASKSSTNA
HGSNSSTSSTSSSSSKSGKGNSGTSTTETITTPLLIDYKKFTPYKDAYQMSNNNFNLSIN
YGSGLLGYWAGILAIAIFANMIKKMFPSLTNNLSGSISNLFRKHLFLPATFRKKKAQEFS
IGVYGFFDGLIPTRLETIIVVIFVVLTGLFSALHIHHVKDNPQYATKNAELGHLIADRTG
ILGTFLIPLLILFGGRNNFLQWLTGWDFATFIMYHRWISRVDVLLIIVHAITFSVSDKAT
GKYKNRMKRDFMIWGTVSTICGGFILFQAMLFFRRKCYEVFFLIHIVLVVFFVVGGYYHL
ESQGYGDFMWAAIAVWAFDRVVRLGRIFFFGARKATVSIKGDDTLKIEVPKPKYWKSVAG
GHAFIHFLKPTLFLQSHPFTFTTTESNDKIVLYAKIKNGITSNIAKYLSPLPGNTATIRV
LVEGPYGEPSSAGRNCKNVVFVAGGNGIPGIYSECVDLAKKSKNQSIKLIWIIRHWKSLS
WFTEELEYLKKTNVQSTIYVTQPQDCSGLECFEHDVSFEKKSDEKDSVESSQYSLISNIK
QGLSHVEFIEGRPDISTQVEQEVKQADGAIGFVTCGHPAMVDELRFAVTQNLNVSKHRVE
YHEQLQTWA
Search with Accession number P78588
http://www.uniprot.org/uniprot/
www.cbs.dtu.dk/services/TMHMM-2.0
www.cbs.dtu.dk/services/TMHMM-2.0
Predicting Post-translational
Modifications
• Post-translational modifications often occur on similar motifs in
different proteins
• PROSITE is a database containing a list of known motifs, each
associated with a function or a post-translational modification
• You can search PROSITE by looking for each motif it contains in
your protein (the server does that for you!)
• PROSITE entries come with an extensive documentation on each
function of the motif
Searching for
PROSITE Patterns
• Search your protein against PROSITE on ExPAsy
• www.expasy.org/tools/scanprosite
• PROSITE motifs are written as patterns
• Short patterns are not very informative by themselves
• They only indicate a possibility
• Combine them with other information to draw a conclusion
• Remember: Not everything is in PROSITE !
www.expasy.org/tools/scanprosite
P12259
www.expasy.org/tools/scanprosite
Interpreting PROSITE Patterns
• Check the pattern function: Is it compatible with the protein?
• Sometimes patterns suggest nonexistent protein features
• For instance : If you find a myristoylation pattern in a prokaryote,
ignore it; prokaryotic proteins have no myristoylation !
• Short patterns are more informative if they are conserved across
homologous sequences
• In that case, you can build a multiple-sequence alignment
• This slide shows an example
Patterns and Domains
• Patterns are usually the most striking feature of the
more general motifs (called domains)
• Domains are less conserved than patterns but usually
longer
• In proteins, domain analysis is gradually replacing
pattern analysis
Protein Domains
• Proteins are usually
made of domains
• A domain is an
autonomous folding
unit
• Domains are more than
50 amino acids long
• It’s common to find
these together:
• A regulatory domain
• A binding domain
• A catalytic domain
Discovering Domains
• Researchers discover domains by
• Comparing proteins that have similar functions
• Aligning those proteins
• Identifying conserved segments
• A domain is a multiple-sequence alignment
formulated as a profile
• For each column, a domain indicates which amino
acid is more likely to occur
Domain Collections
• Scientists have been discovering and characterizing protein
domains for more than 20 years
• 8 collections of domains have been established
• Manual collections are very precise but small
• Automatic collections are very extensive but less informative
• These collections
• Overlap
• Have been assembled by different scientists
• Have different strengths and weaknesses
• We recommend using them all!
The Magnificent 8
• Pfam is the most extensive manual collection
• Pfam is often used as a reference
Searching Domain Collections
• Domains in Pfam often include known functions
• A match between your protein and a domain is desirable
• A match is a potential indication of a function
• This is VERY informative for further research!
• Three servers exist to compare proteins and domain
collections:
• InterProScan
www.ebi.ac.uk/interproscan
• CD-Search (conserved Domain)
www.ncbi.nih.nlm.gov
• Motif Scan
www.ch.embnet.org
Using InterProScan
• InterProScan is the most
comprehensive search engine
for domain databases
• Makes it possible to compare
alternative results on most
collections
• Does not provide a statistical
score
>sp|P53539|FOSB_HUMAN Protein fosB OS=Homo sapiens GN=FOSB PE=1 SV=1
MFQAFPGDYDSGSRCSSSPSAESQYLSSVDSFGSPPTAAASQECAGLGEMPGSFVPTVTA
ITTSQDLQWLVQPTLISSMAQSQGQPLASQPPVVDPYDMPGTSYSTPGMSGYSSGGASGS
GGPSTSGTTSGPGPARPARARPRRPREETLTPEEEEKRRVRRERNKLAAAKCRNRRRELT
DRLQAETDQLEEEKAELESEIAELQKEKERLEFVLVAHKPGCKIPYEEGPGPGPLAEVRD
LPGSAPAKEDGFSWLLPPPPPPPLPFQTSQDAPPNLTASLFTHSEVQVLGDPFPVVNPSY
TSSFVLTCPEVSAFAGAQRTSGSDQPSDPLNSPSLLAL
www.ebi.ac.uk/InterProScan
www.ebi.ac.uk/InterProScan
The CD-Search Output
• CD search is less extensive than that of InterProScan
• Results come with a a statistical evaluation (E-value)
• 10e-15
Low E-value
Good match
• 2.1
High E-value
Bad match
www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi
www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi
www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi
Predicting Functions
with Domains
• Finding a match with a domain having a catalytic function is
good news . . . but what, exactly, does it mean?
• A match indicates that your sequence has the domain
structure . . . but does it also have the function?
• You cannot say before looking into these details:
• Where are the catalytic residues on the domain?
• Does your sequence have the right residues at these positions?
Looking into the Details
• Catalytic residues are normally highly conserved in
domains
• Motif Scan makes it possible to check whether these
important residues are conserved in your sequence
• High bar above 0 = Highly conserved residues
• Green = Your sequence has an expected residue
• Red = Your sequence has an unexpected residue
Looking into the Details (cont’d.)
 R (Arginine) is highly
expected at this position
High bar
Potential active site
 If your protein has an arginine
on this position . . .
Bar is filled with green
Your protein could be active
myhits.isb-sib.ch/cgi-bin/motif_scan
Protein 3D Structure
Primary, Secondary
and Tertiary Structures
• Proteins are made of 20 amino acids
• Proteins are on average 400 amino acids
long
• Protein structure has 3 levels:
• The primary structure is the sequence of a
protein
• The secondary structure is the local structure
• The tertiary structure is the exact position of
each atom on a 3D model
Secondary Structures
• Helix
• Amino acid that twists like a spring
• Beta strand or extended
• Amino acid forms a line without
twisting
• Random coils
• Amino acid with a structure
neither helical nor extended
• Amino-acid loops are usually coils
Guessing the Secondary Structure
of Your Protein
• Secondary structure predictions are good
• If your protein has enough homologues, expect
80% accuracy
• The most accurate secondary structure prediction
server is PSIPRED
PSIPRED Output
• Conf = Confidence
• 9 is the best, 0 the worst
• Pred = Every amino acid is assigned a letter:
• C for coils
• E for extended or beta-strand
• H for helix
>gi|15892329|ref|NP_360043.1| translocation protein TolB [Rickettsia conorii str. Malish 7]
MRNIIYFILSLLFSVTSYALETINIEHGRADPTPIAVNKFDADNSAADVLGHDMVKVISNDLKLSGLFRP
ISAASFIEEKTGIEYKPLFAAWRQINASLLVNGEVKKLESGKFKVSFILWDTLLEKQLAGEMLEVPKNLW
RRAAHKIADKIYEKITGDAGYFDTKIVYVSESSSLPKIKRIALMDYDGANNKYLTNGKSLVLTPRFARSA
DKIFYVSYATKRRVLVYEKDLKTGKESVVGDFPGISFAPRFSPDGRKAVMSIAKNGSTHIYEIDLATKQL
HKLTDGFGINTSPSYSPDGKKIVYNSDRNGVPQLYIMNSDGSDVQRISFGGGSYAAPSWSPRGDYIAFTK
ITKGDGGKTFNIGIMKACPQDDENSERIITSGYLVESPCWSPNGRVIMFAKGWPSSAKAPGKNKIFAIDL
TGHNEREIMTPADASDPEWSGVLN
bioinf.cs.ucl.ac.uk/psipred//?program=psipred
bioinf.cs.ucl.ac.uk/psipred//?program=psipred
bioinf.cs.ucl.ac.uk/psipred//?program=psipred
bioinf.cs.ucl.ac.uk/psipred//?program=psipred
Predicting Other
Secondary Features
• It is also possible to predict these accurately:
•
•
•
•
Transmembrane segments
Solvent accessibility
Globularity
Coiled/coil regions
• All these predictions have an expected accuracy
higher than 70%
Servers
•
•
•
•
www.predictprotein.org
cubic.bioc.columbia.edu/predictprotein
www.sdsc.edu/predicprotein
www.cbi.pku.edu.cn/predictprotein
Predicting 3D Structures
• Predicting 3D structures from sequences only is almost impossible
• The only reliable way to establish the 3D structure of a protein is to
make a real-world experiment in
• X-ray crystallography
• Nuclear magnetic resonance (NMR)
• Structures established this way are conserved in the PDB database
• “The PDB of my protein” is synonymous with “The structure of my
protein”
Retrieving Protein Structures
from PDB
• All PDB entries are 4-letter words!
• 1CRZ, 2BHL . . .
• Sometimes the chain number is added:
• 1CRZA, 1CRZB . . .
• To access all PDB entries, go to www.rcsb.org
• PDB contains 42,000 entries
• PDB contains the structure of 16,000 unique proteins or RNAs
• You can download the coordinates and display the structure
www.rcsb.org
www.rcsb.org
Displaying a PDB Structure
• You can use any of the online
viewers to display the structure
• They will let you rotate the
structure, zoom in and out, or
color it
• PDB files themselves are not
human-readable
Predicting the Structure
of Your Protein
• The bad news:
• It is very hard to predict protein 3D structures
• The good news:
• Similar proteins have similar structures
• If your favorite protein has a homologue with a known structure .
..
• You can do homology modeling
• How?
• Start with a BLAST (more about that in the next slide)
ncbi.nlm.nih.gov/BLAST
ncbi.nlm.nih.gov/BLAST
BLASTing PDB for Structures
• BLAST your protein against
PDB
• If you get a very good hit, it
means PDB contains a
protein similar to yours
• Your protein and this hit
probably have the same
structure
Be Careful!
• Sometimes only one of the domains contained in your protein
has been characterized
• If that’s the case, the PDB will only contain this domain
• Always check the alignments
• Red line = full protein in PDB
• Blue line = one domain only in this entry
Structures and Sequences
• Highly conserved sequences are often important in the structure
• Make a multiple-sequence alignment to identify these important
positions
• Highly conserved positions are either in the core or important for
protein/protein interactions
3D Predictions
• If you want to predict the structure of your protein
automatically, try the Swiss Model
• Swiss Model makes the BLAST for you
• The program does a bit of homology modeling
• The process delivers a new PDB entry
• You can access it at swissmodel.expasy.org
• Swiss Model gives good results for proteins having
homologues in PDB
zhanglab.ccmb.med.umich.edu/I-TASSER/
zhanglab.ccmb.med.umich.edu/I-TASSER/
3D-BLAST
• Use this technique if you have a structure and you
want to find other similar structures
• Use VAST or DALI to look for proteins having the
same 3D shape as yours
• www.eb.ac.uk/dali
• www.ncbi.nlm.nih/vast
3D Movements
• Most proteins need to move to do their job
• Predicting protein movement is possible using
molecular dynamics
• Check out this site: molmolvdb.mbb.yale.edu
• Good molecular dynamics requires extremely powerful
computers
• Don’t expect miracles from standard online resources