6/3/2015burkhard morgenstern, tunis 2007 multiple alignment and motif searching burkhard morgenstern...

04/21/23 Burkhard Morgenstern, Tunis 2007

Multiple Alignment and Motif Searching

Burkhard Morgenstern

Universität Göttingen

Institute of Microbiology and Genetics

Department of Bioinformatics

Tunis, March 2007


Multiple Alignment and Motif Searching

http://www.gobics.de/

burkhard/teaching/tunis_07.php


www.gobics.de/burkhard/teaching/tunis_07.php


Information flow in the cell



Idea:

Sequence -> Structure -> Function



gap between sequence and structure/function data

Lots of data available at the sequence level

Fewer data at the structure and function level


Exponential growth of data bases


Major goal of bioinformatics: close the gap between sequence information and structure/function information

Most important tool for sequence analysis: sequence comparison

Simple approach: dot plot, more advanced approach: sequence alignment


The dot plot


The dot plot

Gibbs and McIntyre (1970)


The dot plot

Y Q E W T Y I V A R E A Q Y E C I V M R E Q Y

Two sequences to be compared


The dot plot

Y Q E W T Y I V A R E A Q Y E C I V M R E Q Y

Comparison matrix


The dot plot

Y Q E W T Y I V A R E A Q Y E C I X V M R E Q Y

Search pairs of identical residues


The dot plot

Y Q E W T Y I V A R E A Q Y E C I X V X M R X E X X X Q X X Y X X

Dot plot: dot (X) for all pairs of identical residues


The dot plot



The dot plot


Homologies as diagonal lines from top-left to bottom-right corner


The dot plot


Inversions as diagonals from bottom left to top right


The dot plot

Y Q E W T Y Q E V R E Y Q E I C I X V X M R Y X X X Q X X X E X X X X

Repeats as parallel diagonals


The dot plot

Y Q E W T Y Q E V R E Y Q E I C I X V X M R Y X X X Q X X X E X X X X


The dot plot

Advantages:

1. Various types of similarity detectable (repeats, inversions)

2. Useful for large-scale analysis

Use filtering for long sequeces: dots represent matching segments instead of matching single residues


The dot plot


Pair-wise sequence alignment

Evolutionary or structurally related sequences:

alignment possible

Sequence homologies represented by inserting gaps



T Y I V A R E Q Y E C I V M R E A Q Y

Two input sequences



T Y I V A R E Q Y E C I V M R E A Q Y

Comparison matrix for two sequences



T Y I V A R E Q Y E C I X V X M R X E X X A X Q X Y X X Dot plot for two sequences



T Y I V A R E Q Y E C I X V X M R X E X X A X Q X Y X X

Similarities in same relative order over entire seqences



T Y I V A R E Q Y E C I X V X M R X E X A Q X Y X

Global alignment of sequences possible



T Y I V A R E Q Y E C X X I X V X M X R X E X A X Q X Y X X

Alignment corresponds to path through comparison matrix



T Y I V A R E Q Y E C X X I X V X M X R X E X A X Q X Y X X

Matches (red), mis-matches (green), gaps (blue)



(global) alignment: write sequences on top of each other, gaps represented by dash symbols



T Y I V A R E Q Y E

C I V M R E A Q Y

Input sequences



T Y I V A R E - Q Y E

C - I V M R E A Q Y –

alignment of input sequences




C - I V M R E A Q Y -

alignment consists matches (red), mismatches (green) and gaps (blue)





Basic task:

Find ‘best’ alignment of two sequences

= alignment that reflects structural and evolutionary relations





Questions:

1. What is a good alignment?

2. How to find the best alignment?





Idea: consider alignment as hypothesis about evolution of sequences.

gaps correspond to insertions/deletions mismatches correspond to substitutions




C - I V M R E A - Q Y

Problem: astronomical number of possible alignments



T Y I V A R E Q Y E

C I - V M R E A Q Y





- C I V M R E A Q Y –


stupid computer has to find out: which alignment is best ??





First (simplified) rules:

1. minimize number of mismatches

2. maximize number of matches





General assumption: sequences not too distantly related.

In this case: mismatches (substitutions) and gaps (insertions/deletions) unlikely

Consequence: good alignment should reduce gaps and mismatches




C I - V M R E A Q Y –







C I - V M R E A Q Y –

Second (simplified) rule:

minimize number of gaps



T Y I V - A R E - Q Y E

C I - V M - R E A Q Y –

Second (simplified) rule:

minimize number of gaps

Parsimony principle: minimize number of evolutionary events



For protein sequences: different degrees of similarity among amino

acids. counting matches/mismatches

oversimplistic



T Y I V

T L V

Protein sequences to be aligned



T Y I V

T L - V

Possible alignment



T Y I V

T - L V

Alternative alignment



T Y I V

T - L V

Some amino acid residues are more similar to each other than others

Therefore: similarity among amino acid residues has to be taken into account.



T Y I V

T - L V

To assess quality of protein alignments:

use similarity scores for amino acids

s(a,b) similarity score for amino acids a and b



Similarity measured by substitution matrices based on substitution probabilities

Important substitution matrices:

PAM (M. Dayhoff) BLOSUM (S. Henikoff / J. Henikoff)



The PAM matrix:

Consider probability pa,b of substitution a → b (or b → a) for amino acids a and b

Define for amino acids a and b similarity score S(a,b) based on probability pa,b

First task: find out pa,b for every pair of amino acids a, b



The PAM matrix:

Use closely related protein families – no alignment problem, no double substitutions

Construct phylogenetic tree with parsimony method

Count substitution frequencies/probabilities Normalize substitution probabilities Extrapolate probabilities for larger

evolutionary distances



Finally: define similarity score

S(a,b) = log (pa,b / qa qb)

qa = (relative) frequency of amino acid a



T Y I V

T - L V

Given a similarity score s(a,b) for pairs of amino acids, define quality score of alignment as:

sum of similarity values s(a,b) of aligned residues

minus gap penalty g for each residue aligned with a gap



T Y I V

T - L V

Example:

Score = s(T,T) + s(I,L) + s (V,V) - g



T Y I V T - L V

Next question: find alignment with best score

Dynamic-programming algorithm finds alignment with best score.

(Needleman and Wunsch, 1970)



T Y I V A R E A Q Y E

- C I V M R E - Q Y –




T Y I V A R E A Q Y E C I X V X M R X E X X Q X Y X X



T Y I V A R E A Q Y E X X C X I X V X M X R X E X X Q X Y X X



T Y I V A R E A Q Y E

- C I V M R E - Q Y –




T W L V - R E A Q I - C I V M R E - H Y



Score of alignment: Sum of similarity values of aligned residues minus gap penatly




Example: S = - g + s(W,C) + s(L,L) + s(V,V) - g + s(R,R) …




T W L V R E A Q Y I X X C X I X V X M X R X E X X H X Y X X




T W L V R E A Q Y I X X C X Alignment corresponds I X to path through V X comparison matrix M X R X E X X H X Y X X




i T W L V R E A Q Y I X X Dynamic programming: C X Calculate scores S(i,j) I X of optimal alignment of V X prefixes up to positions M X i and j. j R X E H Y

T W L V - R - C I V M R



i T W L V R E A Q Y I X X C X S(i,j) can be calculated from I X possible predecessors V X S(i-1,j-1), S(i,j-1), S(i-1,j). M X j R X E H Y




i T W L V R E A Q Y I X X C X Score of optimal path that I X comes from top left = V X M X S(i-1,j-1) + s(R,R) j R X E H Y




i T W L V R E A Q Y I X X C X Score of optimal path that I X comes from above = V X j-1M X S(i,j-1) – g j R X E H Y

T W L V R - - C I V M R



i-1 i T W L V R E A Q Y I X X C X Score of optimal path that I X comes from left = V X M X S(i-1,j) – g j R X X E H Y

T W L - - V R - C I V M R -



i-1 i T W L V R E A Q Y I X X C X Score of optimal path = I X V X Maximum of these three M X values j R X X E H Y

T W L - - V R - C I V M R -



Recursion formula for global alignment:

For sequences x and y

gijS

gjiS

yxsjiS

jiS

ji

)1,(

),1(

),()1,1(

max),(



T W L V R C I V M R E H Y



T W L V R x x x C x x x I x x V x x M x x R x x E x x H x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x M x x R x x E x x H x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x x M x x R x x E x x H x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x x M x x x R x x E x x H x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x x M x x x R x x x E x x H x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x x M x x x R x x x E x x x H x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x x M x x x R x x x E x x x H x x x Y x x Fill matrix from top left to bottom right:



T W L V R x x x C x x x I x x x V x x x M x x x R x x x E x x x H x x x Y x x x Fill matrix from top left to bottom right:



T W L V R x x x x C x x x I x x x V x x x M x x x R x x x E x x x H x x x Y x x x Fill matrix from top left to bottom right:



T W L V R x x x x C x x x x I x x x V x x x M x x x R x x x E x x x H x x x Y x x x Fill matrix from top left to bottom right:



T W L V R x x x x C x x x x I x x x x V x x x M x x x R x x x E x x x H x x x Y x x x Fill matrix from top left to bottom right:



T W L V R x x x x C x x x x I x x x x V x x x x M x x x R x x x E x x x H x x x Y x x x Fill matrix from top left to bottom right:



T W L V R x x x x x x C x x x x x x I x x x x x x V x x x x x x M x x x x x x R x x x x x x E x x x x x x H x x x x x x Y x x x x x x Fill matrix from top left to bottom right:



T W L V R x x x x x x C x x x x x x I x x x x x x V x x x x x x M x x x x x x R x x x x x x E x x x x x x H x x x x x x Y x x x x x x Find optimal alignment by trace-back procedure



T W L V R x x x x x x C x I x V x M x R x E x H x Y x Initial matrix entries?



i

T W L V R

X X

C X Entries S(i,j) scores

I X of optimal alignment of

j V X prefixes up to positions

M i and j.

R

E

H

Y

T W L V

- C I V



i T W L V R j X X X X X C Entries S(i,0) scores I of optimal alignment of V prefix up to positions M i and empty prefix. R E Score = - i* g H Y T W L V - - - -



T W L V R C I V M R E H Y Initial matrix entries: Example, g = 2



T W L V R 0 -2 -4 -6 -8 -10 C -2 I -4 V -6 M -8 R -10 E -12 H -14 Y -16 Initial matrix entries: Example, g = 2


Pair-wise global alignment

T W L V R E A Q Y I X X C X I X V X M X R X E X X F X Y X X

T W L V - R E A Q I - C I V M R E - F Y


Pair-wise global alignment

Computational complexity: how does program run time and memory depend on size of input data?

l1 and l2 length of sequences:Computing time and memory proportional to

l1 * l2

Time and memory complexity = O(l1 * l2)



More realistic gap penalty: affine-linear instead of linear

Penalty for gap of length l:

c0 + (l-1)* c1

c0 = ‘gap-opening penalty’

c0 = ‘gap-extension penalty’


Pair-wise local alignment

So far: global alignment considered: sequences aligned over their entire length.

But: sequences often share only local sequence similarity (conserved genes or domains)

Most important application: database searching



T W L V R E A Q Y I X X C X I X V X M X R X E X X H X Y X X




Problem:

Find pair of segments with maximal alignment score (not necessarily part of optimal global alignment!)

Equivalent: find path starting and ending anywhere in the matrix.




S(i,j) = max { S(i-1,j-i)+s(ai,bj) , S(i-1,j) – g , S(i,j-i) – g }



Recursion formula for local alignment:

S(i,j) = max { 0 , S(i-1,j-i)+s(ai,bj) , S(i-1,j) – g , S(i,j-i) – g }



T W L V R 0 0 0 0 0 0 C 0 I 0 V 0 M 0 R 0 E 0 H 0 Y 0 Initial matrix entries = 0



T W L V R 0 0 0 0 0 0 C 0 0 I 0 V 0 M 0 R 0 E 0 H 0 Y 0 s(C,T) = -2




gijS

gjiS

yxsjiS

jiS

ji

)1,(

),1(

),()1,1(

max),(



Recursion formula for local alignment:

0

)1,(

),1(

),()1,1(

max),(gijS

gjiS

yxsjiS

jiS

ji



For trace-back:

Store positions imax and jmax with

S(imax ,jmax) maximal



Algorithm by Smith and Waterman (1983)

Implementation: e.g. BestFit in GCG package



Complexity: l1 and l2 length of sequences:computing time

and memory proportional to l1 * l2

Time and space complexity = O(l1 * l2)

Too slow for data base searching! Therefore tools like BLAST necessary for

database searching


The Basic Local Alignment Search Tool (BLAST)

New BLAST version (1997)

Two-hit strategy Gapped BLAST Position-Specific Iterative BLAST

(PSI BLAST)



PSI BLAST:

1. search database with standard BLAST

2. take best hits and create multiple alignment

3. calculate profile from multiple alignment

4. search database again with profile as query



profile for sequence family or motif:

table of amino acid/nucleotide frequencies at any position in alignment.



Profile: frequencies of nucleotides at every position.

seq1 A T T G – A T

seq2 C T T G T A G

seq3 A - - G T A T

seq4 A T G G T G T

seq5 A C T G T A C

A 80 0 0 0 0 80 0

T 0 75 75 0 100 0 60

C 20 25 0 0 0 0 20

G 0 0 25 100 0 20 20


Tools for multiple sequence alignment

s1 T Y I M R E A Q Y E S A Q

s2 T C I V M R E A Y E

s3 Y I M Q E V Q Q E R

s4 W R Y I A M R E Q Y E



s1 - T Y I - M R E A Q Y E S A Q

s2 - T C I V M R E A - Y E - - -

s3 - - Y I - M Q E V Q Q E R - -

s4 W R Y I A M R E - Q Y E - - -




s2 - T C I V M R E A - Y E - - -

s3 - - Y I - M Q E V Q Q E R - -

s4 W R Y I A M R E - Q Y E - - -

General information in multiple alignment: Functionally important regions more conserved than

non-functional regions Local sequence conservation indicates functionality!



s1 - T Y I - M R E A Q Y E S A Qs2 - T C I V M R E A - Y E - - -s3 - - Y I - M Q E V Q Q E R - -

s4 W R Y I A M R E - Q Y E - - -

For phylogeny reconstruction: Estimate pairwise distances between sequences

(distance-based methods for tree reconstruction) Estimate evloutionary events in evolution (parsimony

and maximum likelihood methods)




s2 - T C I V M R E A - Y E - - -

s3 - - Y I - M Q E V Q Q E R - -

s4 W R Y I A M R E - Q Y E - - -

Astronomical number of possible alignments!




s2 - T C I V M R E A - - - Y E -

s3 Y I - - - M Q E V Q Q E R - -

s4 W R Y I A M R E - Q Y E - - -

Astronomical number of possible alignments!




s2 - T C I V M R E A - - - Y E -

s3 Y I - - - M Q E V Q Q E R - -

s4 W R Y I A M R E - Q Y E - - -

Computer has to decide: which one is best??



Questions in development of multiple-alignment programs (as in pairwise alignment):

(1) What is a good alignment? → objective function (`score’)

(2) How to find a good alignment? → optimization algorithm

First question far more important !



Traditional Objective functions:

Define Score of alignments as

Sum of individual similarity scores S(a,b) Gap penalties

Needleman-Wunsch scoring system (1970)



Traditional Objective functions

Can be generalized to multiple alignment

(e.g. sum-of-pair score, tree alignment)

Needleman-Wunsch algorithm can also be generalized to multiple alignment, but:

Very time and memory consuming!

-> Heuristics needed


Multiple sequence alignment

First question: how to score multiple alignments?

Possible scoring scheme:

Sum-of-pairs score



Multiple alignment implies pairwise alignments:

1aboA 36 WCEAQt..kngqGWVPSNYITPVN......

1ycsB 39 WWWARl..ndkeGYVPRNLLGLYP......

1pht 51 WLNGYnettgerGDFPGTYVEYIGrkkisp

1ihvA 27 AVVIQd..nsdiKVVPRRKAKIIRd.....

1vie 28 YAVESeahpgsvQIYPVAALERIN......



Multiple alignment implies pairwise alignments:

Use sum of scores of these p.a.

1aboA 36 WCEAQt..kngqGWVPSNYITPVN......

1ycsB 39 WWWARl..ndkeGYVPRNLLGLYP......

1pht 51 WLNGYnettgerGDFPGTYVEYIGrkkisp

1ihvA 27 AVVIQd..nsdiKVVPRRKAKIIRd.....

1vie 28 YAVESeahpgsvQIYPVAALERIN......



Needleman-Wunsch coring scheme can be generalized from pair-wise to multiple alignment



Complexity:

For sequences of length l1 * l2 * l3

O( l1 * l2 * l3 )

For n sequences ( average length l ):

O( ln )

Exponential complexity!



Needleman-Wunsch coring scheme can be generalized from pair-wise to multiple alignment

Optimal solution not feasible:

-> Heuristics necessary


`Progressive´ Alignment

WCEAQTKNGQGWVPSNYITPVN

WWRLNDKEGYVPRNLLGLYP

AVVIQDNSDIKVVPKAKIIRD

YAVESEAHPGSFQPVAALERIN

WLNYNETTGERGDFPGTYVEYIGRKKISP








Guide tree








Idea: align closely related sequences first!




WW--RLNDKEGYVPRNLLGLYP-

AVVIQDNSDIKVVP--KAKIIRD

YAVESEASFQPVAALERIN

WLNYNEERGDFPGTYVEYIGRKKISP

Profile alignment, “once a gap - always a gap”






YAVESEASVQ--PVAALERIN------

WLN-YNEERGDFPGTYVEYIGRKKISP




WCEAQTKNGQGWVPSNYITPVN-



YAVESEASVQ--PVAALERIN------

WLN-YNEERGDFPGTYVEYIGRKKISP




WCEAQTKNGQGWVPSNYITPVN--------

WW--RLNDKEGYVPRNLLGLYP--------

AVVIQDNSDIKVVP--KAKIIRD-------

YAVESEA---SVQ--PVAALERIN------

WLN-YNE---ERGDFPGTYVEYIGRKKISP




“Greedy” algorithm:

Consider partial solution of bigger problem

search best partial solution, fix solution search second-best partial solution that is consistent

with first solution, fix solution Search third-best partial solution … etc.

E.g.: Rucksack-Problem



WCEAQTKNGQGWVPSNYITPVN--------

WW--RLNDKEGYVPRNLLGLYP--------

AVVIQDNSDIKVVP--KAKIIRD-------

YAVESEA---SVQ--PVAALERIN------

WLN-YNE---ERGDFPGTYVEYIGRKKISP




Most important software program:

CLUSTAL W:J. Thompson, T. Gibson, D. Higgins (1994), CLUSTAL

W: improving the sensitivity of progressive multiple sequence alignment … Nuc. Acids. Res. 22, 4673 - 4680

(~ 18.000 citations in the literature)



Problems with traditional approach:

Results depend on gap penalty

Heuristic guide tree determines alignment;

alignment used for phylogeny reconstruction

Algorithm produces global alignments.



Problems with traditional approach:

But:

Many sequence families share only local similarity

E.g. sequences share one conserved motif


Local sequence alignment

Find common motif in sequences; ignore the rest

EYENS

ERYENS

ERYAS



Find common motif in sequences; ignore the rest

E-YENS

ERYENS

ERYA-S



Find common motif in sequences; ignore the rest – Local alignment

E-YENSERYENSERYA-S



Important methods for local multiple alignment:

•PIMA•MEME/MAST

Idea: expectation maximation.



Traditional alignment approaches:

Either global or local methods!


New question: sequence families with multiple local similarities

Neither local nor global methods appliccable


New question: sequence families with multiple local similarities

Alignment possible if order conserved


The DIALIGN approach

Morgenstern, Dress, Werner (1996),PNAS 93, 12098-12103

Combination of global and local methods

Assemble multiple alignment from gap-free local pair-wise alignments (,,fragments“)



atctaatagttaaactcccccgtgcttag

cagtgcgtgtattactaacggttcaatcgcg

caaagagtatcacccctgaattgaataa



atc------taatagttaaactcccccgtgcttag







caaa--gagtatcacccctgaattgaataa





caaa--gagtatcacc----------cctgaattgaataa



atc------taatagttaaactcccccgtgc-ttag

cagtgcgtgtattactaac----------gg-ttcaatcgcg







Consistency!



atc------TAATAGTTAaactccccCGTGC-TTag

cagtgcGTGTATTACTAAc----------GG-TTCAATcgcg

caaa--GAGTATCAcc----------CCTGaaTTGAATaa



Score of an alignment:

Define score of fragment f:

l(f) = length of fs(f) = sum of matches (similarity values)

P(f) = probability to find a fragment with length l(f) and at least s(f) matches in random sequences that have the same length as the input sequences.

Score w(f) = -ln P(f)




Define score of fragment f:

Define score of alignment as

sum of scores of involved fragments

No gap penalty!




Goal in fragment-based alignment approach: find

Consistent collection of fragments with maximum sum of weight scores



atctaatagttaaaccccctcgtgcttagagatccaaaccagtgcgtgtattactaacggttcaatcgcgcacatccgc

Pair-wise alignment:



atctaatagttaaaccccctcgtgcttagagatccaaaccagtgcgtgtattactaacggttcaatcgcgcacatccgc


recursive algorithm finds optimal chain of

fragments.



------atctaatagttaaaccccctcgtgcttag-------agatccaaaccagtgcgtgtattactaac----------ggttcaatcgcgcacatccgc--


recursive algorithm finds optimal chain of

fragments.



------atctaatagttaaaccccctcgtgcttag-------agatccaaaccagtgcgtgtattactaac----------ggttcaatcgcgcacatccgc--

Optimal pairwise alignment: chain of fragments with maximum sum of weights found by dynamic programming:

Standard fragment-chaining algorithm

Space-efficient algorithm



Multiple alignment:






Multiple alignment:



caaccctgaattgaagagtatcacataa

(1) Calculate all optimal pair-wise alignments



Multiple alignment:




(1) Calculate all optimal pair-wise alignments



Fragments from optimal pair-wise alignments might be inconsistent



Fragments from optimal pair-wise alignments might be inconsistent

1. Sort fragments according to scores

2. Include them one-by-one into growing multiple alignment – as long as they are consistent

(greedy algorithm, comparable to knapsack problem)






Consistency problem






Upper and lower bounds for alignable positions



atc------taatagt taaactcccccgtgcttag

Cagtgcgtgtattact aacggttcaatcgcg





atc------taata-----gttaaactcccccgtgcttag

Cagtgcgtgtatta-----ctaacggttcaatcgcg









site x = [i,p] (sequence i, position p)







Calculate upper bound bl(x,i) and lower bound bu(x,i) for each x and sequence i







bl(x,i) and bu(x,i) updated for each new fragment in alignment



Consistency bounds are to be updated for each new fragment that is included in to the growing Alignment

Efficient algorithm

(Abdeddaim and Morgenstern, 2002)



Advantages of segment-based approach:

Program can produce global and local alignments!

Sequence families alignable that cannot be aligned with standard methods



DIALIGN is available

Online at BiBiServ (Bielefeld Bioinformatics Server)

Downloadable UNIX/LINUX executables at BiBiServ

Source code (email to BM)


Program input

Program usage:

> dialign2-2 [options] <input_file>

<input_file> = multi-sequence file in FASTA-format


Program output

DIALIGN 2.2.1 ************* Program code written by Burkhard Morgenstern and Said Abdeddaim e-mail contact: [email protected] Published research assisted by DIALIGN 2 should cite: Burkhard Morgenstern (1999). DIALIGN 2: improvement of the segment-to-segment approach to multiple sequence alignment. Bioinformatics 15, 211 - 218.

For more information, please visit the DIALIGN home page at

http://bibiserv.techfak.uni-bielefeld.de/dialign/

program call: ./dialign2-2 -nt -anc s

Aligned sequences: length: ================== ======= 1) dog_il4 300 2) bla 200 3) blu 200

Average seq. length: 233.3

Please note that only upper-case letters are considered to be aligned.


Program output

Alignment (DIALIGN format): =========================== dog_il4 1 cagg------ ----GTTTGA atctgataca ttgc------ ---------- bla 1 ctga------ ---------- ---------- --------GC CAAGTGGGAA blu 1 ttttgatatg agaaGTGTGA aacaagctat cctatattGC TAAGTGGCAG 0000000000 0000000000 0000000000 0000000011 1111111111 dog_il4 25 ---------- --ATGGCACT GGGGTGAATG AGGCAGGCAG CAGAATGATC bla 17 ggtgtgaata catgggtttc cagtaccttc tgaggtccag agtacc---- blu 51 ccctggcttt ctATGTGCAC AGAATGGGAG GAAAGTGCCT GCTAGTGAGC 0000000000 0000000000 0000000000 0000000000 0000000000 dog_il4 63 GTACTGCAGC CCTGAGCTTC CACTGGCCCA TGTTGGTATC CTTGTATTTT bla 63 ---------- ---------- ---TTTCCCA TGTGCTCCAT GGTGGAATGG blu 101 CAGGGACTCA GAGAGAATGG AGTATAGGGG TCAGGGCat- ---------- 0000000000 0000000000 0009999999 9999999888 8888888888 dog_il4 113 TCCGCCCCTT CCCAGCACca gcattatcct ---GGGATTG GAGAAGGGGG bla 90 ACCACTCCTT CTCAGCACaa caaagcccaa gaaGGTGTTG CGTTCTAGAC blu 140 ---------- ---------- ---------- ---GGGGTGG CCTTAGGCTC 8888888888 8888888800 0000000000 0007777777 7777777777




cagtgcgtgtattactaac----------ggttcaatcgcg




atc------TAATAGTTAaactccccCGTGC-TTag------

cagtgcGTGTATTACTAAc----------GG-TTCAATcgcg

caaa--GAGTATCAcc----------CCTGaaTTGAATaa--


Program output

Alignment (DIALIGN format): =========================== dog_il4 1 cagg------ ----GTTTGA atctgataca ttgc------ ---------- bla 1 ctga------ ---------- ---------- --------GC CAAGTGGGAA blu 1 ttttgatatg agaaGTGTGA aacaagctat cctatattGC TAAGTGGCAG 0000000000 0000000000 0000000000 0000000011 1111111111 dog_il4 25 ---------- --ATGGCACT GGGGTGAATG AGGCAGGCAG CAGAATGATC bla 17 ggtgtgaata catgggtttc cagtaccttc tgaggtccag agtacc---- blu 51 ccctggcttt ctATGTGCAC AGAATGGGAG GAAAGTGCCT GCTAGTGAGC 0000000000 0000000000 0000000000 0000000000 0000000000 dog_il4 63 GTACTGCAGC CCTGAGCTTC CACTGGCCCA TGTTGGTATC CTTGTATTTT bla 63 ---------- ---------- ---TTTCCCA TGTGCTCCAT GGTGGAATGG blu 101 CAGGGACTCA GAGAGAATGG AGTATAGGGG TCAGGGCat- ---------- 0000000000 0000000000 0009999999 9999999888 8888888888 dog_il4 113 TCCGCCCCTT CCCAGCACca gcattatcct ---GGGATTG GAGAAGGGGG bla 90 ACCACTCCTT CTCAGCACaa caaagcccaa gaaGGTGTTG CGTTCTAGAC blu 140 ---------- ---------- ---------- ---GGGGTGG CCTTAGGCTC 8888888888 8888888800 0000000000 0007777777 7777777777


T-COFFEE

C. Notredame, D. Higgins, J. Heringa (2000), T-Coffee: A novel algorithm for multiple sequence alignment, J. Mol. Biol.


T-COFFEE

Problem with “progressive” approaches:

Strictly global alignments

Use only pair-wise comparison


T-COFFEE

Idea: Start with local and global pair-wise alignments (“primary

library” of alignments)

Construct “scondary library” of residues that are indirectly aligned by primary library.

Re-score residue pairs

Construct final alignment with “progressive” method


T-COFFEE

Advantage:

Combination of local and global approaches

Less sensitive against mis-alignments in progressive proceedure


T-COFFEE


T-COFFEE

T-COFFEE and DIALIGN: Less sensitive to spurious pairwise similarities Can handle local homologies better than

CLUSTAL


Most multi-alignment approaches automated, i.e. based on algorithmic rules. Two components:

Objective function: assess alignment quality

Optimization algorithm: find optimal or near-optimal alignment


Fully automated alignment programs necessary f no expert knowledge available if large amounts of data to be analyzed

But: Often no biologically reasonable

results Often additional information about

homologies etc. available


Idea for improved alignment

Use expert knowledge to influence alignment procedure

DIALIGN with user-defined anchor points


Alignment of large genomic sequences

Alignment of large genomic sequences to identify functional elements (phylogenetic footprinting)

Göttgens et al., 2000, 2001, 2002, … Pollard et al., 2004

DIALIGN, MGA, PipMaker, LAGAN, AVID, Mummer, WABA, …



Gene-regulatory sites identified by mulitple sequence alignment (phylogenetic footprinting)


DIALIGN alignment of human and murine genomic sequences


DIALIGN alignment of tomato and Thaliana genomic sequences



DIALIGN used by tracker for phylogenetic footprinting (Prohaska et al., 2004)




Alignment of Hox gene cluster:





DIALIGN able to identify small regulatory elements, but






Entire genes totally mis-aligned






Entire genes totally mis-aligned Reason for mis-alignment: duplications !



The Hox gene cluster:

4 Hox gene clusters in pufferfish. 14 genes, different genes in different clusters!



The Hox gene cluster:

Complete mis-alignment of entire genes!


Alignment of sequence duplications

S1

S2



S1

S2

Conserved motivs; no similarity outside motifs



S1

S2

Duplication in two sequences



S1

S2

Mis-alignment would have lower score!



S1

S2

Duplication in one sequence



S1

S2


Possible mis-alignment



S1

S2


S3



S1

S2

Consistency problem

S3



S1

S2

More plausible alignment – and higher score:

S3



S1

S2

Consistency problem

S3



S1

S2

Alternative alignment; probably biologically wrong;lower numerical score!

S3


Anchored sequence alignment

Biologically meaningful alignment not possible by automated approaches.

Idea: use expert knowledge to guide alignment procedure

User defines a set anchor points that are to be „respected“ by the alignment procedure



NLFVALYDFVASGDNTLSITKGEKLRVLGYNHN

IIHREDKGVIYALWDYEPQNDDELPMKEGDCMT





Use known homology as anchor point



NLFV ALYDFVASGDNTLSITKGEKLRVLGYNHN


Use known homology as anchor point

Anchor point = anchored fragment (gap-free pair of segments)

Remainder of sequences aligned automatically





Alignment of anchored positions a and b not enforced – a and b may be un-aligned –, but:

a is only residue that can be aligned to b

Residues left of a aligned with residues left of b



-------NLF VALYDFVASG DNTLSITKGE klrvlgynhn

iihredkGVI YALWDYEPQN DDELPMKEGD cmt-------

Anchored alignment





GYQYRALYDYKKEREEDIDLHLGDILTVNKGSLVALGFS

Anchor points in multiple alignment




IIHREDKGVIYALWDYEPQND DELPMKEGDCMT

GYQYRALYDYKKEREEDIDLHLGDILTVNKGSLVALGFS

Anchor points in multiple alignment



-------NLF V-ALYDFVAS GD-------- NTLSITKGEk lrvLGYNhn

iihredkGVI Y-ALWDYEPQ ND-------- DELPMKEGDC MT-------

-------GYQ YrALYDYKKE REedidlhlg DILTVNKGSL VA-LGFS--

Anchored multiple alignment


Algorithmic questions

Goal:

Find optimal alignment (=consistent set of fragments) under costraints given by user-specified anchor points!


Additional input file with anchor points:

1 3 215 231 5 4.5

2 3 34 78 23 1.23

1 4 317 402 8 8.5




NLFVALYDFVASGDNTLSITKGEKLRVLGYNHN IIHREDKGVIYALWDYEPQNDDELPMKEGDCMTGYQYRALYDYKKEREEDIDLHLGDILTVNKGSLVALGFS



1 3 215 231 5 4.5

2 3 34 78 23 1.23

1 4 317 402 8 8.5




1 3 215 231 5 4.5

2 3 34 78 23 1.23

1 4 317 402 8 8.5

Sequences




1 3 215 231 5 4.5

2 3 34 78 23 1.23

1 4 317 402 8 8.5

Sequences start positions




1 3 215 231 5 4.5

2 3 34 78 23 1.23

1 4 317 402 8 8.5

Sequences start positions length




1 3 215 231 5 4.5

2 3 34 78 23 1.23

1 4 317 402 8 8.5

Sequences start positions length score




Requirements:

Anchor points need to be consistent! – if necessary: select consistent subset from user-specified anchor points






Inconsistent anchor points!



atctaat---agttaaactcccccgtgcttag

Cagtgcgtgtattac-taacggttcaatcgcg


Inconsistent anchor points!



Requirements:

Anchor points need to be consistent! – if necessary: select consistent subset from user-specified anchor points

Find alignment under constraints given by anchor points!



Use data structures from multiple alignment






Greedy procedure for multiple alignment






Question: which positions are still alignable ?



atctaatagttaaactcccccgtgcttag Si


caaagagtatcacccctgaattgaataa x

For each position x and each sequence Si exist an

upper bound ub(x,i) and a lower bound lb(x,i) for

residues y in Si that are alignable with x






ub(x,i) and lb(x,i) updated during greedy procedure






Initial values of lb(x,i), ub(x,i)






ub(x,i) and lb(x,i) updated during greedy procedure



Anchor points treated like fragments in greedy algorithm:

Sorted according to user-defined scores Accepted if consistent with previously accepted

anchors

ub(x,i) and lb(x,i) updated during greedy

procedure

Resulting values of ub(x,i) and lb(x,i) used as initial

values for alignment procedure






Initial values of lb(x,i), ub(x,i)






Initial values of lb(x,i), ub(x,i) calculated using anchor

points



Ranking of anchor points to prioritize anchor points, e.g.

anchor points from verified homologies -- higher priority

automatically created anchor points (using CHAOS, BLAST, … ) -- lower priority


Application: Hox gene cluster



Use gene boundaries as anchor points



Use gene boundaries as anchor points

+ CHAOS / BLAST hits



no anchoring anchoring

Ali. Columns

2 seq 2958 3674

3 seq 668 1091

4 seq 244 195

Score 1166 1007

CPU time 4:22 0:19



Example:

Teleost Hox gene cluster:

Score of anchored alignment 15 % higher than score of non-anchored alignment !

Conclusion: Greedy optimization algorithm does a bad job!


Application: Improvement of Alignment programs

Two possible reasons for mis-alignments:

Wrong objective function: Biologically correct

alignment gets bad numerical score

Bad optimization algorithms: Biologically correct

alignment gets best numerical score, but algorithm

fails to find this alignment


Application: Improvement of Alignment programs

Two possible reasons for mis-alignments:

Anchored alignments can help to decide


Application: RNA alignment



aa----CCCC AGC---GUAa gucgcuaucc a

cacucuCCCA AGC---GGAG Aac------- -

ccg----CCA AaagauGGCG Acuuga---- -

non-anchored alignment



aa----CCCC AGC---GUAa gucgcuaucc a

cacucuCCCA AGC---GGAG Aac------- -

ccg----CCA AaagauGGCG Acuuga---- -

structural motif mis-aligned



aaCCCCAGCG UAAGUCGCUA UCca--

--CACUCUCC CAAGCGGAGA AC----

----CCGCCA AAAGAUGGCG ACuuga

3 conserved nucleotides as anchor points


WWW interface at GOBICS(Göttingen Bioinformatics Compute Server)


WWW interface at GOBICS (Göttingen Bioinformatics Compute Server)


Gene predictions for eukaryotes



Goal: find location and structure of protein-coding genes in eukaryotic genome sequences.



attgccagtacgtagctagctacacgtatgctattacggatctgtagcttagcgtatctgtatgctgttagctgtacgtacgtatttttctagagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctagagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcttagtcgtgtagtcttgatctacgtacgtatttttctagagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctagagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgtacgtacgtatttttctagagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctagagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctagtcgtagtcgtagtcgttagcatctgtatggtcgtagtcgttagcatctgtatgctgttagctgtacgtacgtatttttctaggggagcttcgtagtctatggctag



Three different approaches to computational gene-finding:

Intrinsic: use statistical information about known genes (Hidden Markov Models)

Extrinsic: compare genomic sequence with known proteins / genes

Cross-species sequence comparison: search for similarities among genomes


Hidden-Markov-Models (HMM) for gene prediction

Generative probabilistic model for sequence of observations („symbols“).

Finite set of states

States can emit symbols Transitions between states possible Sequence generated by path between states



Example: The occasionally dishonest casino.

3 5 6 6 6 4 6 5 1 6 5 1 2

F F U U U U U F F F F F F

Possible states:

fair (F); unfair (U); begin (B); end (E)



Assumptions:

Emission probabilities known; depend only on current state.

Transition probabilities known, depend only on current state



F

U

E B



3 5 6 6 6 4 6 5 1 6 5 1 2 s

B F F U U U U U F F F F F F E φ

For sequence s and parse φ:

P(φ) probability of φ P(φ,s) joint probability of φ and s = P(φ) * P(s|φ) P(φ|s) a-posteriori probability of φ



3 5 6 6 6 4 6 5 1 6 5 1 2

B F F U U U U U F F F F F F E

Goal: find path φ with maximum a-posteriori probability P(φ|s)

Idea: find path that maximizes joint probability P(φ,s) by dynamic programming (Viterbi algorithm)



Application to gene prediction:

A T A A T G C C T A G T C s (DNA) Z Z Z E E E E E E I I I I φ (parse)

Introns, exons etc modeled as states in GHMM („generalized HMM“)

Given sequence s, find parse that maximizes P(φ|s)

(S. Karlin and C. Burge, 1997)


AUGUSTUS

Basic model for GHMM-based intrinsic gene finding comparable to GenScan (M. Stanke)


AUGUSTUS


AUGUSTUS

Features of AUGUSTUS:

Intron length model Initial pattern for exons Similarity-based weighting for splice sites Interpolated HMM Internal 3’ content model



A T A A T G C C T A G T C s (DNA) Z Z Z E E E E I I I I φ (parse)

Explicit intron length model computationally expensive.


AUGUSTUS

Intron length model:

• Explicit length distribution for short introns• Geometric tail for long introns

Intron (fixed)

Exon

Intron (expl.)

Exon

Intron (geo.)


AUGUSTUS


AUGUSTUS

Extension of AUGUSTUS using include extrinsic information:

Protein sequences EST sequences Syntenic genomic sequences User-defined constraints


Gene prediction by phylogenetic footprinting

Comparison of genomic sequences

(human and mouse)


catcatatcttatcttacgttaactcccccgt

cagtgcgtgatagcccatatccgg





Standard score:Consider length, # matches, compute probability of random occurrence



Translation option:





Translation option:

L S Y V

catcatatc tta tct tac gtt aactcccccgt

cagtgcgtg ata gcc cat atc cgg

I A H I

DNA segments translated to peptide segments; fragment score based on peptide similarity:

Calculate probability of finding a fragment of the same length with (at least) the same sum of BLOSUM values



P-fragment (in both orientations)

L S Y V



I A H I

N-fragment catcatatc ttatcttacgtt aactcccccgtgct || | | | cagtgcgtg atagcccatatc cg

For each fragment f three probability values calculated; Score of f based on smallest P value.



P-fragment (in both orientations)

L S Y V



I A H I

N-fragment catcatatc ttatcttacgtt aactcccccgtgct || | | | cagtgcgtg atagcccatatc cg

P-fragments associated with strand and reading frame!




AGenDA: Alignment-based Gene Detection Algorithm



Fragments in DIALIGN alignment



Build cluster of fragments



Identify conserved splice sites



•Candidate exons bounded by conserved splice sites •Find optimal chain of candidate exons



0%10%20%30%40%50%60%70%80%90%

100%

sensitivity specificity

AGenDAGenScan



AGenDA

GenScan

64 %

12 % 17 %


AUGUSTUS+

Extended GHMM using extrinsic information

Additional input data: collection h of `hints’ about possible gene structure φ for sequence s

Consider s, φ and h result of random process. Define probability P(s,h,φ)

Find parse φ that maximizes P(φ|s,h) for given s and h.


AUGUSTUS+

Hints created using

Alignments to EST sequences Alignments to protein sequences Combined EST and protein alignment (EST

alignments supported by protein alignments) Alignments of genomic sequences User-defined hints


AUGUSTUS+

Alignment to EST: hint to (partial) exon

EST

G1


AUGUSTUS+

EST alignment supported by protein: hint to exon (part), start codon

EST

G1

Protein


AUGUSTUS+

Alignment to ESTs, Proteins: hints to introns, exons

ESTs, Protein

G1


AUGUSTUS+

Alignment of genomic sequences: hint to (partial) exon

G2

G1


AUGUSTUS+

Consider different types of hints:

type of hints: start, stop, dss, ass, exonpart, exon, introns

Hint associated with position i in s (exons etc. associated with right end position) max. one hint of each type allowed per position in s Each hint associated with a grade g that indicates its source or reliability.


AUGUSTUS+

hi,t = information about hint of type t at position i

hi,t = $ if no hint of type t available at i

hi,t = [grade, strand, (length, reading frame)] if hint available

(hints created by protein alignments or DIALIGN contain information about reading frame)


AUGUSTUS+

Standard program version, without hints

A T A A T G C C T A G T C s (sequence) Z Z Z E E E E E E I I I I φ (parse)

Find parse that maximizes P(φ|s)


AUGUSTUS+

AUGUSTUS+ using hints

A T A A T G C C T A G T C s (sequence) $ $ $ $ $ $ $ X $ $ $ $ $ h (type 1) $ $ $ $ $ $ $ $ $ $ $ $ $ h (type 2) $ $ $ $ X $ $ $ $ $ $ $ $ h (type 3) . . . .

Z Z Z E E E E E E I I I I φ (parse)

Find parse that maximizes P(φ|s,h)


AUGUSTUS+

As in standard HMM theory: maximize joint probability P(φ,s,h)

How to define P(φ,s,h) ?


AUGUSTUS+

General assumption: Hints of different types t and at different positions i independent of each other (for redundant hints: ignore „weaker“ types).


AUGUSTUS+


),|(),(),,( shPsPhsP


AUGUSTUS+


),|(),(),,( shPsPhsP

ti

ti shPshPsP,

, ),|(),|(),(


AUGUSTUS+

Assumption: P(hi,t |φ,s) depends on type t, grade g and whether hi,t is compatible with φ or s.

Example: hi,t hint to exon E

hi,t compatible with parse φ if E part of φ.

hi,t compatible with sequence s if start and stop codons exist according to E and if no internal stop codon in E exists


AUGUSTUS+

For given g and t: 3 possible values for P(hi,t |φ,s)

P(hi,t |φ,s) = q+(t,g) if hi,t compatible with φ

P(hi,t |φ,s) = q-(t,g) if hi,t compatible with s

but not compatible with φP(hi,t |φ,s) = 0 if hi,t not compatible with s

Values learned from training data


AUGUSTUS+

Results:

Gene (sub-)structures supported by hints receive bonus compared to non-supported structures

Gene (sub-)structures not supported by hints receive malus

(M. Stanke et al. 2006, BMC Bioinformatics)


AUGUSTUS+


AUGUSTUS+

h, h’ collections of hints;

h’i,t = hi,t for (i,t) ≠ (I,T)

h’I,T ≠ hI,T = $; g grade of h’I,T

φ+, φ- gene structures on s

h’IT compatible with φ+, but not with φ-


AUGUSTUS+

),|'(),(

),|'(),(

)',,(

)',,(

)',|(

)',|(

shPsP

shPsP

hsP

hsP

hsP

hsP


AUGUSTUS+

),|'(),(

),|'(),(

)',,(

)',,(

)',|(

)',|(

shPsP

shPsP

hsP

hsP

hsP

hsP

titi

titi

shPsP

shPsP

,,

,,

),|'(),(

),|'(),(


AUGUSTUS+

),|'(),(

),|'(),(

)',,(

)',,(

)',|(

)',|(

shPsP

shPsP

hsP

hsP

hsP

hsP

titi

titi

shPsP

shPsP

,,

,,

),|'(),(

),|'(),(

ti TI

TIti

TI

TI

titi

shP

shPshPsP

shP

shPshPsP

, ,

,,

,

,

,,

),|(

),|'(),|(),(

),|(

),|'(),|(),(


AUGUSTUS+

ti TI

TIti

TI

TI

titi

shP

shPshPsP

shP

shPshPsP

, ,

,,

,

,

,,

),|(

),|'(),|(),(

),|(

),|'(),|(),(


AUGUSTUS+

ti TI

TIti

TI

TI

titi

shP

shPshPsP

shP

shPshPsP

, ,

,,

,

,

,,

),|(

),|'(),|(),(

),|(

),|'(),|(),(

),|$(

),(),|(),(

),|$(

),(),|(),(

,

,

shP

gTqshPsP

shP

gTqshPsP

TI

TI


AUGUSTUS+

),|$(

),(),|(),(

),|$(

),(),|(),(

,

,

shP

gTqshPsP

shP

gTqshPsP

TI

TI


AUGUSTUS+

),|$(

),(),|(),(

),|$(

),(),|(),(

,

,

shP

gTqshPsP

shP

gTqshPsP

TI

TI

),|$(),(

),|$(),(

),|(

),|(

,

,

shPgTq

shPgTq

hsP

hsP

TI

TI


AUGUSTUS+

Result:

i.e. structure φ+, which is compatible with additional hint h’IT receives relative bonus

),|$(),(

),|$(),(

,

,

shPgTq

shPgTq

TI

TI

),|(),(

),|(),(

),|(

),|(

),|(

),|(

,

,

shPgTq

shPgTq

hsP

hsP

hsP

hsP

TI

TI


AUGUSTUS+

Results (gene level) on data set sag178

% SN % SP

Augustus 42 38

GenScan 18 14

GeneID 17 17

HMMGene 20 7

Aug. + EST 49 46

Aug. + prot 71 68

Aug. combined 68 65

Aug. all 82 79

GenomeScan 37 38

TwinScan 20 25


AUGUSTUS+

Using hints from DIALIGN alignments:

1. Obtain large human/mouse sequence pairs (up to 50kb) from UCSC

2. Run CHAOS to find anchor points3. Run DIALIGN using CHAOS anchor points4. Create hints h from DIALIGN fragments5. Run AUGUSTUS with hints


AUGUSTUS+

Hints from DIALIGN fragments:

Segment covered by peptide fragment minus 33 bp at both ends defines exon part hint on all 6 reading frames.


AUGUSTUS+

Hints from DIALIGN fragments:

Consider fragments with score ≥ 20

Distinguish high scores (≥ 45) from low scores Consider reading frame given by DIALIGN Consider strand given by DIALIGN

=> 2*2*2 = 8 grades


AUGUSTUS+

AUGUSTUS best ab-initio method at EGASP


EGASP test results

AUGUSTUS

GENSCAN

geneid GeneMark.hmm

Genezilla

0

10

20

30

40

50

60

70

80

90

100 Nukleotid Level

Sensitivität

Spezifität


EGASP test results

AUGUSTUS

GENSCAN

geneid GeneMark.hmm

Genezilla

0

10

20

30

40

50

60

70

80

90

100 Exon Level

Sensitivität

Spezifität


EGASP test results

AUGUSTUS

GENSCAN

geneid GeneMark.hmm

Genezilla

0

2,5

5

7,5

10

12,5

15

17,5

20

22,5

25

27,5

30 Transkript Level

Sensitivität

Spezifität


EGASP test results

AUGUSTUS

GENSCAN

geneid GeneMark.hmm

Genezilla

0

2,5

5

7,5

10

12,5

15

17,5

20

22,5

25

27,5

30 Gen Level

Sensitivität

Spezifität


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Sn Sp Sn Sp Sn Sp Sn Sp

Base Exon Transcript Gene

Ac

cu

rac

y

AUGUSTUS

AUGUSTUS+DIALIGN

DOGFISH-C

SGP2

TWINSCAN

TWINSCAN-MARS

N-SCAN

EGASP test results


Ongoing projects

Brugia malayi (TIGR)

Aedes aegypti (TIGR)

Schistosoma mansoni (TIGR)

Tetrahymena thermophilia (TIGR)

Galdieria Sulphuraria (Michigan State Univ.)

Coprinus cinereus (Univ. Göttingen)

Tribolium castaneum (Univ. Göttingen)

6/3/2015burkhard morgenstern, tunis 2007 multiple alignment and motif searching burkhard morgenstern...

Documents

q y e c i x v x

r e q y e c i v

dot plot y q e w t y

dot plot burkhard morgenstern

right burkhard morgenstern

cellburkhard morgenstern

functionburkhard morgenstern

gapsburkhard morgenstern