|
|
|
|
 |
|
| |
|
| Hinxton,
Friday 2 December 2005 |
|
What to sequence next Pick one species at a time |
|
|
|
![Fabio Pardi [left] and Nick Goldman.](../../../../../images/press/press05/press25nov05ebipics.jpg) |
| Fabio Pardi [left] and Nick Goldman. |
|
Press
Release 2 December 2005 [PDF]
After humans, mice, chickens and others what genomes
should scientists sequence next? In a paper published
today in PLoS Genetics, Fabio Pardi and Nick
Goldman of the EMBL-European Bioinformatics Institute
present a way to decide. Surprisingly, they show
that always choosing the next best single species
is just as effective as planning to sequence several
genomes in advance.
DNA sequencing has revealed a vast amount
of information about biology. But genome sequencing
remains expensive and time consuming, so scientists
need a strategy to help them select the organisms
that will give them the most new information.
One
solution is to sequence the most distantly related
organisms, to get the widest possible diversity
of sequences. Biologists represent the relationships
between different species as a tree, with the length
of the branches varying according to the degree
by which their DNA sequences differ. "If we are
prepared to assume that the most informative set
is the one with the greatest evolutionary divergence,
the problem of which species to sequence next can
be solved by observing the length of the branches
that separate the unsequenced species from those
that have already had their genomes sequenced, and
choosing the organism that's separated from the
others by the longest sequence of branches", explains
Fabio Pardi.
The tendency has been for centres to
choose a group of new genomes to sequence. However,
the current study shows that picking the best candidates
one at a time is equally informative. "Computer
scientists call this a 'greedy strategy' because
it involves always taking the best bet for yourself",
says Nick Goldman. "However, if, say, a centre had
enough funding to sequence five organisms, we might
expect to get a better set of genomes by considering
all five together. Counterintuitively, we found
that in this case the greedy strategy is the best.
We were surprised because in computer science greed
is definitely not good – greedy algorithms seldom
provide the best solution to a problem."
"Our findings have clear implications for planning
large-scale genome sequencing efforts", continues
Pardi. "Provided that they remain open about their
choices so that two different sequencing centres
don't choose the same genome, selecting the next
most attractive organism to sequence is just as
effective as having a long-term strategy."
Evolutionary divergence isn't
the only factor that scientists consider when choosing
which genomes to sequence, but other criteria can
be factored into Goldman and Pardi's greedy strategy
so long as those criteria can be quantified. For
example, sequencing costs, or the economic importance
of an organism, could be considered. Their strategy
can also be applied to different problems, such
as conservation biology. "Of course, we're not advocating
that genome scientists or conservation biologists
stop working cooperatively, but at least they can
feel confident about sequencing or conserving the
organism of their choice without messing things
up for their collaborators," says Goldman.
Source article
Species choice for comparative genomics: being greedy works
F. Pardi and N. Goldman
PLoS Genet., 25 November 2005
Press contact
Cath Brooksbank PhD
EMBL-EBI Scientific Outreach Officer Wellcome Trust
Genome Campus, Hinxton, Cambridge CB10 1SD, UK
Tel: +44 [0]1223 492525
E-mail:
cath@ebi.ac.uk
Sarah Sherwood
EMBL Information Officer, European Molecular Biology
Laboratory, Meyerhofstrasse 1, 69117 Heidelberg,
Germany
Tel: +49 [0] 6221 387-125
E-mail: sarah.sherwood@embl.de |
|
|
|
|
