Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Loss of information at deeper divergences
David Penny Phylomania Nov 2013
The mathematicos caused the problem!!! Now they should solve it!
Okay, maybe we could help them, Here are some ideas
Need relative – not absolute - information
the comfort zone
ML Int
ML Rel
Mlav ML
MLan MP ML
MLep MP MP
popn classic phylogeny deep
phylogeny
can we go further back
in time?
Markov models - Loss of information
damned eukaryotes!
Calculated results, Δ ≤ ¼ + ne-qt
-0.2
0
0.2
0.4
0.6
0.8
1
1 10 100 1000 10000
0.01 0.005 0.002 0.001
0%
20%
40%
60%
80%
100%
120%
0.1 1 10
pe
rce
nta
ge
of
tre
es
co
rre
ct
d=0.001
d=0.100
d=0.500
d=1.000
d=2.000
d=5.000
infinite
1. simulation results with covarion model
number of internal edges correct, out of 6neighbor joining, 9 taxa, 1000 columns, i.i.d.
00.5
1
5 8 13 20 32 50 80 125
200
320
500
790
1250
2000
millions of years (log scale)
6
5
4
3
2
1
0
simulation results with standard model
1 idea, delete fast sites
If there were a mixture of a) faster evolving sites, and b) and we could identify them c) and remove them would that help go further back in time?
deleting faster sites
Ancestral Sequence Reconstruction
Giardia animals plants
2, 3, testing
Ancestral Sequence Reconstr-
uction
vaults 3-D info
subgroups X and Y
a b c d e k l m n o
ax a
y
subgroup X subgroup Y
4. gene length vs similarity
f1 a . . . . a . . . . . .
f2 . . . . . a . . . . a .
f3 . . . . . a . . . . . a
f4 . . . . . a . . . . . .
g1 . a a a a . a a . . . .
g2 a a a a a . . a . . a .
h1 . . a a a . . . a a . .
h2 . . a . a . . . a a . .
h3 a . . a a . . . a a . .
i j 3 4 5 6 7 8 9 0 1 2
i j i j
. . . a
a . a a
upper bound = 17
lower bound = 12
?
13
Would weighting by incompatibilities
help?
5, Weighting
information from sequence order not used
Alignment Reordered Alignment original sequence order shuffled/reordered AIIFLNSALGPSPELFPIILATKVL ASAGPSPPATPLLIIIILLFFNEKV
AIMFLNSALGPPTELFPVILATKVL ASAGPPTPATPLLIMVILLFFNEKV
SIMFLNHTLNPTPELFPIILATETL SHTNPTPPATPLLIMIILLFFNEET
TILFLNSSLGLQPEVTPTVLATKTL TSSGLQPPATPLLILTVLVTFNEKT
TLLFLNSMLKPPSELFPIILATKTL TSMKPPSPATPLLLLIILLFFNEKT
ALLFLNSTLNPPTELFPLILATKTL ASTNPPTPATPLLLLLILLFFNEKT
AILFLNSFLNPPKEFFPIILATKIL ASFNPPKPATPLLILIILFFFNEKI
c! ways to reorder alignment
shuffle by columns & by taxa
6. So, could we use ‘words’ of 2, 3, 4, 5, … letters
7. Alphabet reduction