comparison Pairwise table Calculate all pairwise alignment scores and arrange them in a table S1 S2 S3 S4 S5 2 0 9 1
S1 10 5 4 S2 10 25 8 S3 5 25 11 S4 4 8 11 S5 2 0 9 1
Convert all score into distances ... 1. FengDoolitle : D=log(SSrand)/(SmaxSrand) 2. Model based distances log det formular maximum likelihood
root S9 S6 S8
S9
S2 S3
S4 S1 S5 S6 S2 S3 S5 S7 S7
S8
S4
S1
Progressive Profile Alignment
Build Profiles
freeze alignments
optimal profile alignment
ClustalW Thompson Higgins Gibson 1994 Put together some good ideas:
Distances of pairs of sequences are based on a full stochastic model of sequence evolution (... to be discussed soon) The guide tree is computed by a valid method of phylogenetic tree reconstruction Saitou and Nei 1987 The multiple alignment is put together by progressive profile alignment Sequence weighting is applied Gaussian branching processes Altschul Carroll Lipman 1989
Contributions from biology:
Score matrices influence the alignment results. The score matrix used to score pairwise alignments is chosen on the basis of the evolutionary distance of the sequences. Different matrices for closely related sequences and remote pairs of sequences.
The hydrophobic core of a protein is more conserved than its surface. Position specific gapopen profile penalties are multiplied by a modifier that is a function of the residues observed at the position. Hydrophobic residues (which are more likely to be in the well conserved core of the protein) give higher gap penalties than hydrophilc residues (which are more likely to be on the water accesible and less conserved surface of the protein)
Loops on the surface of a protein are often missing in other members of a protein family.
Gap open penalties are also decreased if the position is spanned by a consecutive strech of five or more hydrophilic residues.
Insertions and deletions are rare events, but once they occur, they are propagated and show up in many family members at the same position.
Both gapopen and gapextend penalties are increased if there are no gaps in a column but gaps occur nearby in the partial alignments. This tries to enforce that gaps show up at the same position.
A pairwise alignment whispers, a full multiple alignment shouts out loud. In the progressive alignment stage, if the score of a profile alignment is low, the guide tree may be adjusted on the fly to defer the low scoring alignment until later when more profile information has been accumulated.
...along the same lines: Early decissions might be wrong since there was little profile information available at this time. Remove the first sequence from the alignment and realign it to the almost full alignment using sequence to profile alignment. Continue with the second, third ,... sequence. Iterate for some time. (not implemented in ClustalW)
6 & 2
s r q p p %D ` & ` Aai0 h e% $ 2 # 0 ) ) 7 1 ) h D D B ) 1 # ) H0 ` 4 %5$ 6 D 4 & b D 4 2 D g b # ) D D f 2 ) b & 4 D AeG5$ US4 A# G0 5eG6 %" ! $ #
b b D 1 0 4 2 D c 4 0 b 0 ' & ' # T & XA5$ d I Ga(A5$ ` 2 D ) $ 7 # 30 WYAX1 W@0 V US3I HG# 4 ) $ B ) ) T R Q Q P E D D B 2 # 0 4 1 7 6 # $ 4 2 1 0 & F) CA@98530 ) (' & %" ! $ #
0.9 0.9 0.02
A
0.02 0.06
T
0.06 0.06
Purines A,G Pyrimidines C,T
0. 02
0.06
0.
02 0. 02
0.02
G
0.02 0.9
C
0.9
No Action 0.9 Transition 0.06 Transversion 0.02
0. 02
SS G S S G S G
0) U8W85g ) D b ' & 0 $ & 4 # ) U2 0 5%US4 X9AD (%e5%# V AaW%D ` D $ D f D 1 2 $ D g D $ 4 2 D 0 $ D f 2 0 # D 7 2 0 4 2 # D US4 '3GG6 U3AX1 (& %(S4 a%3) & 4 & f D 7 ) # !
& b D D $
"
!
"
!
D U' X 0 A%YG5$ 4 f 2 # 0 4 & 0 b & D
E GU8W85g ) D b ' & 0 $ & # 2 G6 & $ V # GX9AU%) US4 ' AW ) D 1 2 D 7 D D f c 2 D g 0 2 D f ) ) ) D 1 # AUS4 0 GXY5$ ` ) (iY) Uf A# ) 1 0 4 & f 1 # 4 D 2 # 6 2& B 6 AXY) US4 B %& %UA5 0 %(S4 2 # 1 D D f 6 D 4 $ D 2 D ) 4 & f $ %D44eDb W@0 V US4 eUAG6 a7 e5 GXY5$ ` 4 ) $ D f D 4 # 2 D ) 4 D ) ) D 1 # 1 0 4 & f 1 # 4 ) (iY) & GU30 V G6 AaW%D ` D af ) D 2 D 4 2 D 0 $ ) 0
D U' 4 # D # 4 & f 4 & f ) 4@1YD ` D (& G6 %(& B " ! ) %(S4 6 & ! ) %(S4 U (& A ! 2 4 & f & # 2 D V
D # (& G6 D # (& G6
4 & f %(& B !
D U' 4 # 4 1 ) @YD ` D ) %(S4 U (& A 4 & f & # 2 D V
!
D U' 4 # 4 1 ) @YD ` D 4 & %(& f B " ! ) %(S4 U (& A ! 4 & f & # 2 D V
E aAGU7 a# ) 2 # 0 4 ) D D
)GD))GDX1Y5$ d e( 8& 1 58& %D ` 5$ ` # g # $ 6 D b b & D $ c 4 $ # ) 0 f f 4 0 ) D ) ) D 1 # aS4 S9& GGXY5$ ` ) (iY4 ! 1 0 4 & f 1 #
4 ) & ` US4 V # D f
4 2 D ) D 0AG5$ ` US4 AW D f 2 D g 0 42ADG6 2AD ` G6 30 0 5YS7 V Uf D 2 ) D $ 7 4 D E (%(& %USe# 30 $ ! ) 6 $ # $ D f 4 2
2 A#
6 D 2 0 ) D T UX GG6 ) 0 AaW%D ` D US4 %& US4 0 %(f I 4 2 D 0 $ D f c & D f ) 4 & ) GaXWA# 5YS7 V ) D # 1 4 7 D $ 7 4 DX9AUAb V 30 U2 eAa& %(f ) 1 2 D 7 2 # c b D 4 7 b # ) ' D g & )GD%5) ) & ` & AUS4 ) T %@G& 8(i1 I 4 & 4 4 b b 2 D f $ D 4 1 $ & f D 4 & 4 4 2 D ) D %5) AG5$ ` US4 U (& V 0 B 7 ! D f & # 2 D 4
4 ) ) D D f ) 0 Gb US4 0
5YS7 V US4 @YD V & D $ 7 4 D f ) 4 1 V D # 1 4 7 2 $ D 6 b D aXWA# & %G# Uf !
) D # 1 4 7 GaXWA# 5YS7 V A# A%9%# V 30 D $ 7 4 2 2 # 0 4 & $ 2 2 & 4 2 # ) D # 1 4 7 4 30 5AX1 GaXWA# ) & ` B & %UA5 2 ! b $ D 2 D
30 (f 2 & )GD4%&5) 4 &
g # $ 6 D b b & ) ) ) D 1 # D f e( 8& 1 0 GXY5$ ` US4 $ D ' 7 D b ' & 4 2 7 # $ D 4 0 2 V # %UG%2 U5AX1 %# a30 V ) 4 ) 0 ) 2 # D 1 D 4 & 4 D f V V # aAX1 XG& ` ) %5) US4 A !
XG& D 1 6 D b b & ) ) ) D 1 # 1 0 GXY5$ d e( g # DU3G# ` V # e) US4 & b ' 0 ) ) 4 D D f b ) D 4 & 4 4 2 D $ V 0 B E G%5) A5%D V (6 ) ) D 1 # GXY5$ ` US4 AaW%D ` D f 4 2 D 0 $
) 4 0 ) D 4 & 4 G%5) %# B $ D U' UAX1 6 b 7 #
` ) %5) D 4 & 4 $ 8& & V # $ D 2 D %UA5 2 $ 7 YA# V 30 2 D US4 2 ! D f
GGa(A5$ ` ) D 0 4 0 b 0 ' & ' # 2A#040a& 4 1 58& GGa(A5$ ` ) 2 $ 6 D b b & D $ ) D 0 4 0 b 0 ' & ' # D ) D GUf ! ) ! ) ) D 0 4 0 b 0 ' & ' # GGa(A5$ ` (A(6 AX1 b & 2 # 0 4 0 2 # D f c 2 D g 0 $ ) US4 ' AWS6 0 AaW%D ` D Uf ! 4 2 D 0 $ D
Uf ! 6 b # 6 & " ! 2 ) D 0 4 0 b 0 ' & ' # GGa(A5$ ` 58& GWAD US4 X930 ! D $ ) D 0 $ 4 2 D f D 1 2
2 # 0 4 0 ) 2 $ 6 D b b & ) Aa& 4 1 0
W%9 0 $ 4 & 0 $ 4 & ) 0 W%9 af
! ) !
)
!
D $ D 5%Uf $ ! W%9 & 30 GGa(A5$ ` 0 $ 4 & 2 ) D 0 4 0 b 0 ' & ' # 2A#040a& 4 & GW89 %3) 6 & ' ) 2 $ b b D 0 $ & 7 2 ) $ D ' 7 c ) D 4 & 4 D f D 4 $ D 7 2 W%UG%2 ' G%5) US4 %& %a%AD D (& B a30 V 0 XG& ` ) %5) US4 A ! D 4 0 2 ) D 1 D 4 & 4 D f V 2 & 1
US4 V # & %# V GGa(A5$ ` D f b b $ ) D 0 4 0 b 0 ' & ' # b & 2 # 0 4 0 2 # D & D f D ) (A(6 AX1 a 8) US4 a7 ' 4 7 B ` D4) fi1GD %# V AaW%D ` D A5%D V (6 & $ 4 2 D 0 $ 4 2 D $ V 0 ` 7 5S6 eU2 G6 D $ (Uf & D $ 4 # # ) 6 b #
&
! !
!
! ) !
)
6 & 2
& b b
$ %# V
4 & f D 7 ) %(S4 a%3) & D $ !
D g & D D %(f (& b ` & D
$ 7 YA#
2 !
A3W(6 85) 2 # 0...