What is Ka/Ks?

The rati o of the number of nonsynonymous

substi tuti ons per nonsynonymous si te (Ka)
to the number of synonymous
substi tuti ons per synonymous si te (Ks).
Sorry,what was that?
Let us start from the begi nni ng. I magi ne
you al i gn the sequences of the same gene
from two speci es. There wi l l usual l y be
di fferences between the sequences
(evol uti on!). Some of these wi l l l ead to
di fferences i n the ami no aci ds of the
encoded protei n (nonsynonymous changes)
and some, because of the degeneracy of the
geneti c code, l eave the protei n unchanged
(synonymous, or si l ent changes). Counti ng
up the number of each gi ves us a measure
of the amount of change of the sequence.
Then we have to adjust these fi gures.
I see,we now have measures of the rates of
evolution.So why adjust anything?
Due to the degenerate nature of the code,
onl y about 25% of the possi bl e changes i n
our sequence are synonymous. I magi ne
that our gene i s not under sel ecti on i n
ei ther speci es; that i s, i t i s evol vi ng
neutral l y wi th chance al one determi ni ng
whether a new mutati on goes from rare to
common. Most such mutati ons are l ost by
chance, but the chance that a gi ven new
neutral mutati on wi l l go to fi xati on i n a
di pl oi d popul ati on i s 1/2N, where N i s the
popul ati on si ze. I n thi s case, the l i kel i hood
that a nonsynonymous mutati on woul d go
to fi xati on i s the same as that for a
synonymous mutati on. So, i f I correct for
the degeneracy of the code, I shoul d have a
method that reports that the number of
nonsynonymous changes at each possi bl e
nonsynonymous si te i s the same as the
number of synonymous changes per
synonymous si te; that i s, Ka/Ks = 1.
Devi ati ons from a rati o of one wi l l then tel l
me somethi ng about the sel ecti ve forces
acti ng on the protei n, gi ven that Ks i s
tel l i ng me the background rate of evol uti on.
Sounds simple enough
Unfortunatel y not. For exampl e, consi der
the codons speci fyi ng asparti c aci d and
l ysi ne: both start AA, l ysi ne ends A or G,
and asparti c aci d ends T or C. So, i f the
rate at whi ch C changes to T i s hi gher
from the rate that C changes to G or A (as
i s often the case), then more of the changes
at the thi rd posi ti on wi l l be synonymous
than mi ght be expected. Many of the
methods to cal cul ate Ka and Ks di ffer i n
the way they make the correcti on needed
to take account of thi s bi as.
Doesnt the time that has lapsed since the
two species separated matter?
Good point. As sequences diverge over time,
the observed number of changes
underestimates the real number of changes.
I magine a nucleotide that begins as A.
I n one lineage it is replaced by a C and then
again by a T, the two changes would still
only cause one difference in the alignment.
Also, sites that match in the alignment
could have changed independently, but
ended up the same. Fortunately, the extent
of real di vergence can be esti mated from
the total observed amount of di vergence.
These mul ti -hi t correcti onmethods al so
di ffer i n thei r sophi sti cati on. However,
none can work mi racl es: as the number of
changes i ncreases, the amount of
i nformati on from the al i gnment decreases
and we approach saturati on, i n whi ch case
the data are usel ess.
I can see this is getting a bit complicated.
Does all this really matter?
Unfortunatel y i t often does. For exampl e,
the rel ati onshi p between GC content and
Ks i s i mportant i n the debate about the
evol uti on of GC content i n humans, but
the pattern i s very sensi ti ve to the method
of esti mati on.
OK,so I have Ka and Ks for my gene in a
given comparison between species.So what?
Wel l , you now have fi gures for the amount
of protei n evol uti on (Ka) that i s
comparabl e between genes from the same
speci es compari son. I f sel ecti on does not
act on si l ent si tes (a bi g i f!), from the
neutral theory of evol uti on, Ks shoul d al so
be proporti onal to the mutati on rate of the
gene. Thi s i s because, whi l e the
probabi l i ty that a new neutral mutati on
goes to fi xati on i s 1/2N, they are created
each generati on at a rate 2N, where i s
the neutral mutati on rate per generati on.
Hence, the rate of neutral evol uti on i s
2N/2N = and i t i s thi s rate that Ks i s
measuri ng. I f our method has gone to
pl an, the rati o of the Ka and Ks al so
now tel l s us about the way the gene
has evol ved.
What we usual l y fi nd i s that Ka i s
much l ess than Ks (i .e. Ka/Ks << 1; Fi g. 1)
because a mutati on that changes a
protei n i s much l ess l i kel y to be di fferent
between two speci es than one whi ch i s
si l ent; that i s, most of the ti me sel ecti on
el i mi nates del eteri ous mutati ons,
keepi ng the protei n as i t i s (puri fyi ng
sel ecti on).
Sort of what you expect,given that proteins
evolved to do what they do and do it
ratherwell.
Yes, but thi s i s not al ways the case.
I n a few i nstances (often when i mmune
system genes co-evol ve wi th parasi tes),
we fi nd that Ka i s much greater than Ks
(i .e. Ka/Ks >>1). Thi s i s strong evi dence
that sel ecti on has acted to change the
protei n (posi ti ve sel ecti on).
I get it.So if Ka equals Ks then evolution of
the sequence must be neutral?
Not so fast. Neutral evol uti on i s a
possi bi l i ty that cannot be excl uded. But,
what i f one part of the gene (one protei n
domai n, say) was under posi ti ve sel ecti on,
but other parts under puri fyi ng sel ecti on?
Then you mi ght get an average Ka/Ks = 1.
There are recent methods that al l ow you
to take a mul ti pl e sequence al i gnment,
the phyl ogeny of the speci es i nvol ved and
work out a Ka/Ks rati o for each codon. I f i n
al l l i neages a codon i s undergoi ng posi ti ve
sel ecti on, thi s i s a powerful method to
TRENDS in Genetics Vol.18 No.9 September 2002
http://tig.trends.com 0168-9525/02/$ see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(02)02722-1
486 Forum
Back to Basics
The Ka/Ks ratio:diagnosing the form of sequence evolution
TRENDS in Genetics
0
.
0
5
0
.
1
5
0
.
2
5
0
.
3
5
0
.
4
5
0
.
5
5
0
.
6
5
0
.
7
5
0
.
8
5
0
.
9
5
1
.
0
0
0.1
0
0.2
0.3
0.4
0.5
0.6
Ka/Ks
F
r
e
q
u
e
n
c
y
Fig. 1. The frequency of different values of Ka/Ks for
835 mouserat orthologous genes. Figures on the
xaxis represent the middle figure of each bin; that is,
the 0.05 bin collects data from 0 to 0.1
detect i t. Al ternati vel y, you can ask
whether there i s a di fferent rati o i n one
l i neage, suggesti ng that somethi ng
happened pecul i ar to that speci es. These
new sorts of anal yses are reveal i ng much
more posi ti ve sel ecti on than we suspected.
Forget the details,how can I convert my
alignment into Ka and Ks estimates?
There are several di fferent ways you
coul d do thi s. There i s an excel l ent free
package for PCs cal l ed MEGA2 that
i mpl ements l oads of di fferent methods:
http://www.megasoftware.net/. I f you have
GCG on a UNI X machi ne, the di verge
package cal cul ates the Li 93 protocol (not
the best, but OK). The codon-based or
l i neage-based methods are i mpl emented
i n PAML: http://abacus.gene.ucl .ac.uk/
software/paml .html . Thi s wi l l al so
cal cul ate an ol d measure (Nei and
Gojobori ) and a new esti mate to the
maxi mum-l i kel i hood method (Yang and
Ni el sen). PAML can be i mpl emented on
Uni x, l i nux, PCs and Macs (OS X
and before).
Where can I read more?
Nei , M. and Kumar, S. (2000)
Molecular Evolution and Phylogenetics,
Oxford Uni versi ty Press. The book that
goes wi th MEGA2.
Yang, Z.H. and Bi el awski , J.P. (2000)
Stati sti cal methods for detecti ng mol ecul ar
adaptati on. Trends Ecol. Evol. 15, 496503
Laurence D.Hurst
Dept of Biology and Biochemistry,
University of Bath, Bath, UK BA2 7AY.
e-mail: bssldh@bath.ac.uk
TRENDS in Genetics Vol.18 No.9 September 2002
http://tig.trends.com 0168-9525/02/$ see front matter 2002 Elsevier Science Ltd. All rights reserved.
487 Forum
Book Review
Science,a social
product?
The Perversion of Knowledge:The True
Story of Soviet Science
by Vadim J . Birstein
Westview, 2001. US$ 32.50 (512 pages)
ISBN 0 8133 3907 3
Thi s i s a
di squi eti ng book.
Vadi m J. Bi rstei n
recounts the story
of sovi et sci ence
(~19201960) by
l i sti ng the fates of
sovi et sci enti sts
and thei r
persecutors. The
story begi ns i n the
twenti es, when sovi et pol i ti ci ans tri ed to
create a new type of sci ence, a sci ence that
was no l onger done by the sons of the
bourgeoi si e, but by the sons and daughters
of the worki ng cl ass. For them, the cl ass to
whi ch a sci enti st bel onged was more
i mportant than thei r work. The pol i ti ci ans
di d not care whether sci ence was ri ght or
wrong, merel y that i t fol l owed or cl ai med
to fol l ow party pol i cy. The party and the
secret pol i ce were acti ve i n enforci ng thi s
pol i cy, and many sci enti sts who spoke out
agai nst i t were shot. One of the defenders
of the ol d-fashi onedsci ence was Nobel
pri ze wi nner Pawl ow (18491936), famous
for studyi ng refl exes i n dogs. The ol d man
was fearl ess; i n 1934 he wrote to Buchari n,
one of the organi zers of revol uti on who by
then had l ost al l power, My Lord, i t has
become so hard for any decent person to
l i ve i n your soci al i st paradi se.However,
Pawl ow was too famous for the secret
servi ce to act agai nst hi m.
Next, Birstein concentrates on the
attack of Lysenko and his supporters on
biological science. The general story is well
known. Lysenko, a peasant, maintained
that he could educateplants, heritably
altering their characteristics. He posed as a
saviour, claiming to increase agrarian
production gigantically. Yet it was all fraud.
I t is deeply moving to read of those who did
not accept Lysenkos lies; most had to leave
science and many were murdered. The fates
of the plant breeder Vavilov and others are
described in detail. Some objectors survived;
for example, I osif Rapoport, an excellent
geneticist who had shown that certain
chemicals are mutagenic in Drosophila.
I n the World War I I , he lost an eye and
received the highest medal a Jewish officer
could receive in the Red Army. I n August
1948, he publicly called one of the defenders
of Lysenko an obscurantistat the notorious
meeting of the Academy of Agricultural
Sciences. He lost his job, but he survived. I t
is unsettling to read that Alexander Oparin,
well known for his ideas on the origin of life,
praised Lysenko.
Throughout the book one aspect i s
mi ssi ng the actual research done by the
sci enti sts. The reader woul d l i ke to know
thei r rel evant di scoveri es. I nstead, we
l earn a mass of i nteresti ng new detai l s
about the persecuti ons i n the Lysenko
affai r. The fate of sci enti sts i s not the
enti re hi story of sci ence. The sci enti fi c
thoughts, experi ments and breakthroughs
shoul d al so be documented, al though thi s
i s certai nl y more l abori ous.
But Bi rstei n has al so di scovered
somethi ng essenti al l y new: a bi ochemi st,
Gri gory Mai ranowsky, used pol i ti cal
pri soners under sentence of death to test
vari ous poi sons. Mai ranowsky began hi s
human experi ments i n 1935, al though the
l ab books of hi s experi ments have
di sappeared, so one can onl y guess whi ch
poi sons he and hi s col l aborators tested.
I n any case, the experi ments seem rather
si mi l ar to the experi ments done i n German
concentrati on camps; for exampl e, those by
Professor Hi rt i n the concentrati on camp of
Natzwei l er. Mai ranovsky was jai l ed from
1951 to 1961, but l ater hi s career
recovered, and he became di rector of a
provi nci al i nsti tute. What a story.
I t destroys the i l l usi on that that
murderous experi ments wi th humans
were the speci al i ty of the Germans or
Japanese; the Sovi ets di d them as wel l .
Genetics was in worse shape in the
Soviet Union than in Nazi Germany, where
genetics was only partially demolished.
I n the Soviet Union, genetics was almost
completely destroyed. Truth had completely
evaporated. I n this context the author
makes an interesting comparison.
He compares the work of the GESTAPO
(the German secret service) with the work of
the GPU/NKVD (the Soviet secret service).
The former tried to get the truth about the
enemies of Nazism to annihilate all of them.
The latter did not care about truth,
just wanting signatures on prefabricated
confessions. Both institutions worked
differently, but were equally despicable.
I n retrospect, i t i s astoni shi ng that
bi ol ogi cal sci ence i n the USSR survi ved to
begi n agai n i n the si xti es. Apparentl y, i t i s
di ffi cul t to destroy sci ence: i t mi ght al most
be dead, but i t wi l l recover. What do we
l earn from the book? Truth i s the essence
of sci ence. Secrecy and l i es si gnal i ts i l l