1. Reverse-­‐engineering
techniques
In
Data
Integra3on
David
Gomez-­‐Cabrero

2. The
Unit
COMP
MODEL
WET-­‐LAB
BIOINFORMATICS
MEDICAL
INFORMATICS
More
than
30
researchers
EVERYTHING
CONNECTED

3. The
Unit

4. What
is
Reverse
Engineering?
Open
Piece
1:
The
quesMons
connecMons
Piece
3:
The
Piece
2:
The
Data
dynamics
IntegraMon

5. What
is
Reverse
Engineering?

6. REVERSE
SYSTEMS
ENGINEERING
BIOLOGY
DEFINE
DEFINE
MATH
HYPOTHESIS
ELEMENTS
INTERACTION
DESCRIPTION
PIECE
1
PIECE
2

7. REVERSE
ENGINEERING

8. “…the
process
of
analyzing
a
subject
system
to
iden%fy
its
components
and
their
rela7onships
and
to
create
representa7ons
of
the
system
in
another
form
or
at
a
higher
level
of
abstrac7on…
Generally
involves
extrac7ng
design
ar7facts
and
building
or
synthesizing
abstrac7ons
that
are
less
implementa7on
dependent…”
E.
J.
Chikofsky
and
J.
H.
Cross,
“Reverse
Engineering
and
Design
Recovery:
A
toxonomy”,
IEEE
SoKware,
vol
7,
no.
1,
1990,
pp
13-­‐17.

9. Requirements
Design
Source
Code
Behaviour

10. Behaviour

11. Requirements
Behaviour
Design
Source
Code
Design
Behaviour

12. LETTERS
Decoding the ancient Greek astronomical calculator
known as the Antikythera Mechanism
T. Freeth1,2, Y. Bitsakis3,5, X. Moussas3, J. H. Seiradakis4, A. Tselikas5, H. Mangou6, M. Zafeiropoulou6, R. Hadland7,
D. Bate7, A. Ramsey7, M. Allen7, A. Crawley7, P. Hockley7, T. Malzbender8, D. Gelb8, W. Ambrisco9 & M. G. Edmunds1
The Antikythera Mechanism is a unique Greek geared device, con- planetary cycles. We note that a major aim of this investigation is
structed around the end of the second century BC. It is known1–9 to set up a data archive to allow non-invasive future research, and
that it calculated and displayed celestial information, particularly access to this will start in 2007. Details will be available on www.an-
cycles such as the phases of the moon and a luni-solar calendar. tikythera-mechanism.gr.
Calendars were important to ancient societies10 for timing agricul- The back door inscription mixes mechanical terms about con-
tural activity and fixing religious festivals. Eclipses and planetary struction (‘‘trunnions’’, ‘‘gnomon’’, ‘‘perforations’’) with astronom-
motions were often interpreted as omens, while the calm regular- ical periods. Of the periods, 223 is the Saros eclipse cycle (see Box 1
ity of the astronomical cycles must have been philosophically for a brief explanation of astronomical cycles and periods). We
attractive in an uncertain and violent world. Named after its place discover the inscription ‘‘spiral divided into 235 sections’’, which is
of discovery in 1901 in a Roman shipwreck, the Antikythera
Mechanism is technically more complex than any known device
for at least a millennium afterwards. Its specific functions have
remained controversial11–14 because its gears and the inscriptions
upon its faces are only fragmentary. Here we report surface
imaging and high-resolution X-ray tomography of the surviving
fragments, enabling us to reconstruct the gear function and double
the number of deciphered inscriptions. The mechanism predicted
lunar and solar eclipses on the basis of Babylonian arithmetic-
progression cycles. The inscriptions support suggestions of mech-
anical display of planetary positions9,14,15, now lost. In the second
century BC, Hipparchos developed a theory to explain the irregu-
larities of the Moon’s motion across the sky caused by its elliptic
orbit. We find a mechanical realization of this theory in the gear-
ing of the mechanism, revealing an unexpected degree of technical
sophistication for the period.
The bronze mechanism (Fig. 1), probably hand-driven, was ori-
ginally housed in a wooden-framed case1 of (uncertain) overall size
315 3 190 3 100 mm (Fig. 2). It had front and back doors, with
astronomical inscriptions covering much of the exterior of the mech-
anism. Our new transcriptions and translations of the Greek texts are Figure 1 | The surviving fragments of the Antikythera Mechanism. The 82
fragments that survive in the National Archaeological Museum in Athens are
given in Supplementary Note 2 (‘glyphs and inscriptions’). The
shown to scale. A key and dimensions are provided in Supplementary Note 1
detailed form of the lettering can be dated to the second half of the (‘fragments’). The major fragments A, B, C, D are across the top, starting at
second century BC, implying that the mechanism was constructed top left, with E, F, G immediately below them. 27 hand-cut bronze gears are
during the period 150–100 BC, slightly earlier than previously sug-

13. Goals:
-­‐
as
an
approach
to
study
the
design
or
-­‐
as
a
prerequisite
for
re-­‐design.

14. REVERSE
SYSTEMS
ENGINEERING
ENGINEERING
BIOLOGY

15. coupled to ef®cient ribosome-bin
W. C. Mechanism and regulation of eukaryotic protein synthesis. Microbiol. Rev. 56, 291± becauseC.,B. formation. Proc. Natl Acad. Sci.Puri®cation and characterization of its components. Such coope
initiation complex
13. Merrick, W.
of stochastic ¯uctuations
Michael Kemper, W. M. & Anderson, W. F. USA 69, 3602±3605 (1972). of homogenous repression (low `leakiness'),
Elowitz & Stanislas Leibler in
). `rational M2ASchreier, M. H.reticulocytes. J. Biol. Chem. 250, 5556±5562 (1975). II.to the comparable proteinof mRNA
initiation factor network design' may lead both
from rabbit and engineering and IP
V., Borukhov, S. I. & Hellen, C. U. T. Eukaryotic ribosomes require initiation factors 1 and 14. Trachsel, H., Emi, B., & Staehelin, T. Initiation of mammalian protein synthesis. The
te initiation codons. Nature 394, 854±859 (1998).
new cellular complex with puri®edBiology and Physics, Princetongeneralabout the values of paramete
Departments of behaviours and to an116, 755±767 (1977). tainty obstacle to the of ofp
assembly of the initiation Molecular initiation factors. J. Mol. Biol. improved understanding University, Princeton, design
15. Benne, R., Brown-Luedi, M. L. & Hershey, J. W. B. Puri®cation and characterization of protein
ti, A. & Maitra, U. Functions of eukaryotic initiation factor 5 in the formation of an 80S naturally 08544, USA eIF-4D, and eIF-5 from rabbit reticulocytes. J. Biol. Chem. tions between different componen
New Jersey occurring networks.
synthesis initiation factors eIF-1, eIF-4C,
re
253, 3070±3077 (1978).
polypeptide chain initiation complex. J. Biol. Chem. 266, 14039±14045 (1991). In the network shown in initiation factor in the ®rst 80S the order of magnitude of the in
16. Peterson, D. T., Safer, B. & Merrick, W. C. Role of eukaryoticFig. 1a, 5theformation ofrepressor protein, LacI
..............................................................................................................................................
re
hesevich, J. & Maitra, U. Molecular cloning and expression of cDNA for mammalian initiation complexes. J. Biol. Chem. 254, 7730±7735 (1979). compatible with the possibility o
from E. coli,I.inhibits Jackson, R. J. & Hellen, C. U. T. A prokaryotic-like mode of out repressor gene, arti®
Fletcher, S. P., the transcription of the second many essential
Networks ofN., interactinginitiation codon during internal initiation of increase the chances that the d
17. Pestova, T. V., Shatsky, biomolecules carry
initiation factor 5. Proc. Natl Acad. Sci. USA 90, 3058±3062 (1993).
Forward
engineering
-K., Yoon, H., Hannig, E. M. & Donahue, T. F. GTP hydrolysis controls stringent selection of tetR from theliving cells1,virus RNAs. Genes Dev. 12,transposon the oscillatory regime, we made tw
functions in tetracycline-resistance 67±83 (1998).
binding of cytoplasmic eukaryotic ribosomes to the
translation of Hepatitis C and Classical Swine fever but the `design principles' underlying the
Tn10, whose pro-
nents. First, to address transcripti ¯
tart codon during translation initiation in Saccharomyces cerevisiae. Genes Dev. 11, 2396± tein product in turn inhibits the expression of remaingene, yet from l
functioning of such intracellular networks a third strong, cI tightly repressibl
used poorly under- n
7).
ENGINEERING
SyntheMc
Biology
, Lee, J. H., Zoll, W. L., Merrick, W. C. & Dever, T. E. Promotion of Met-tRNAMet binding to
Acknowledgements
phage. despite intensive efforts expression,quantitative which 6cycle. the m
Finally, CI inhibitsand R. Schneider for antibodies, and previously, analysis of l
stood, Merrick for discussions, D. Etchison
We thank W.
lacI including completing the combine
by yIF2, a bacterial IF2 homolog in yeast. Science 280, 1757±1760 (1998).
That suchfor simple systems2.were supportedwelead to temporalsequencesthat of mRNAla
relativelyandsequencing eIF5B.feedback loop cangrants from the lifetimes closer to
L. Siconol®-Baez a negative These studies
Here by present a 7 complementary
operator
oscillations to brin
. Second,
NIH to C.U.T.H. T.V.P.
etters to nature
Choi, S. K., Roll-Mecak, A., Burley, S. K. & Dever, T. E. Universal conservation in translation
revealed by human and archaeal homologs of bacterial translation factor IF2. Proc. Natl
USA 96, 4342±4347 (1999).
in the concentrations of each of its components can be8 insertedfrom aa ¯
approach to this problem: theto T.V.P.
Correspondence and requests for materials should be addressed
simple model of transcriptional regulation, which we used to design the att
(e-mail: tpestova@netmail.hscbklyn.edu).
synthetic network to implement a particular recognize this tag andused m
design and sequence ,seen 39 end of each rep
coli ), we
construction of
at the
function. We target
a carboxy-termi
the repressilator and study its possible behaviours (BoxpartInshown to redu
three transcriptional repressor systems that are not 1). of any e Such tags have been this
C., Dessen, P., Hershey, J. W. B., Plumbridge, J. A. & Grunberg-Manago, M. Sequence of the
model, the action clock3±5 to build an oscillating network,factors, more
natural biological of the network depends on several termed of
DNA-binding domain from
actor IF2 gene; unusual protein features and homologies with elongation factors. Proc. Natl ................................................................. (ref. 8) and diminishcon- e
including the dependence of transcription rate to about100 min (ref. 11). on repressor
the half-life
Reporter theoscillatoryrate, and 8C.decay leastof the protein in mi
A synthetic translation network At rates these considerations 30 the individua
USA 81, 7787±7791 (1984). 30±40
a Repressilator
y, D., Dewey, K. F., Hershey, J. W. B. & Thach, R. E. Guanosine 59-triphosphatase activity of centration, With
andtranscriptionalDepending colonies were tracked manua
NATURE | VOL 403 | 20 JANUARY 2000 | www.nature.com biology techniques to construct a
actor f2. Proc. Natl Acad. Sci. USA 61, 1066±1070 (1968).
PLlac01
Colburn, T. et al. Light-scattering studies showing the effect of initiation factors on the
of messenger RNA. regulators the values ofrepressilator and a compatible, hig on these parameters,
dissociation of Escherichia coli ribosomes. J. Mol. Biol. 94, 461±478 (1975). at least two types of solutions are quanti®ed.
Michael B. Elowitz & Stanislas Leibler
possible: the system the tet-repressible tainingmay converge promot
intermediate stability variant of gfp
ampR
et al. In vitro study of two dominant inhibitory GTPase mutants of Escherichia coli toward a stable steady state, or the steady state interferes with repression by L IPTG may become
tetR-lite
initiation factor IF2. Direct evidence that GTP hydrolysis is necessary for factor recycling. leading to sustained limit-cycle oscillations IPTG 1b, the ¯u
unstable, Molecular Biology and Physics, Princeton University, Princeton, pulse of(Fig. mightc). capable of
Departments of
New Jersey 08544, USA
The timecourse of be
em. 274, 6074±6079 (1999).
, A. H., Sarkar, P. & Maitra, U. Release of polypeptide chain initiation factor IF-2 during
We foundtet01 oscillations are favouredTemporal plasmids and grow
PL that Fig. 2. by repressilator-containing cells. A cul
.............................................................................................................................................. ing the two oscillation
strong promoters
kanR coupledofto ef®cient ribosome-binding essential tight transcriptional be a si
interacting biomolecules carry out many sites,
functions in living cells1, but the `design principles' underlying increase in cell after transfer overall the displayed what appeared to ¯uoresce
complex formation. Proc. Natl Acad. Sci. USA 69, 3602±3605 (1972). Networks
TetR
W. C., Kemper, W. M. & Anderson, W. F. Puri®cation and characterization of homogenous repression such intracellular networks remain poorly repressionshown). per
TetR of (low `leakiness'), cooperative under- ¯uorescence Because individual ce characteristics,
actor M2A from rabbit reticulocytes. J. Biol. Chem. 250, 5556±5562 (1975).
pSC101
functioning
gfp-aav
and comparable protein and mRNA decay rates maintainingFig. 1b). A we s
stood, despite intensive efforts including quantitative analysis of (Box 1, synchronization, 150 minutes, roughly threefo not
!PR
H., Emi, B., Schreier, M. H. & Staehelin, T. Initiation of mammalian protein synthesis. II. The 2
f theorigin complex with puri®ed initiation factors. J. Mol. Biol. 116, 755±767 (1977).
initiation
general simple systems . the design and time. The amplitude under oscil
relatively
obstacle to Here design ofconstruction of a networks intensity as they mic
approach to this problem: the
we present a complementary lating single cells
biochemical ¯uorescence is uncer-the gre of
! cI LacI tainty about theimplementofparticular function. We characterize the consisting of hund
synthetic network to values a parameters that used microcolonies interac-
three between repressor components. In of any GFP ¯uorescence. levels of experiments, total observation tim
Brown-Luedi, M. L. & Hershey, J. W. B. Puri®cation and characterization of protein
GFP
nitiation factors eIF-1, eIF-4C, eIF-4D, and eIF-5 from rabbit reticulocytes. J. Biol. Chem. tionstranscriptionaldifferentsystems that arenetwork, our network, estimates phase after
natural biological clock to build an oscillating 3±5
not part
termed entering a stationary
of
±3077 (1978).
the order of magnitude of the exhibit parameters seem behaviou relevant oscillatory to be
lacI-lite
D. T., Safer, B. & Merrick, W. C. Role of eukaryotic initiation factor 5 in the formation of 80S
complexes. J. Biol. Chem. 254, 7730±7735 (1979). compatible with the possibility determined Nevertheless, to
NATURE | VOL 403 | 20 JANUARY 2000 | www.nature.com
ColE1 of oscillations. by a Fourier a
V., Shatsky, I. N., Fletcher, S. P., Jackson, R. J. & Hellen, C. U. T. A prokaryotic-like mode of
! cI-lite
increase the chances that the arti®cial network would function in
cytoplasmic eukaryotic ribosomes to the initiation codon during internal initiation of the oscillatory regime, we made two alterations to naturalas estimate range of periods, compo-
of Hepatitis C and Classical Swine fever virus RNAs. Genes Dev. 12, 67±83 (1998).
PLtet01 nents. First, to address transcriptional strength and tightness, min (m intervals, is 160 6 40 we
used strong, yet tightly repressible hybrid promoters, developed GFP levels in the two siblin
dgements previously, which combine the l PL promoter with lac and tet
W. Merrick for discussions, D. Etchison and R. Schneider for antibodies, and operator sequences6. Second, to bring the effective repressor protein one another for long perio
Baez for sequencing eIF5B. These studies were supported by grants from the

16. L
We thank W. Merrick for discussions, D. Etchison and R. Schneider for antibodies, and
R operator sequences6. Second, to br
L. Siconol®-Baez for sequencing eIF5B. These studies amp supported by grants from the
were
tetR-lite lifetimes closer to that of mRNA
Box 1 NIH to C.U.T.H. and T.V.P.
coli7), we insertedtet01
PL a carboxy-term
Network design for materials should be addressed to T.V.P.
Correspondence and requests
TetR
kanR
sequence8, at the 39 end of each re
(e-mail: tpestova@netmail.hscbklyn.edu). TetR
recognize this tag and target the at
pSC101 gfp-aav
!PR Such tags have been shown to redu
origin
Design of the repressilator started cI ! with a simple mathematical modelfrom mor
LacI DNA-binding domain of
GFP
.................................................................
transcriptional regulation. We did not set out to describe precisely thehalf-life o (ref. 8) and diminish the
to about 30±40 min (ref. 11).
Forward
engineering
lacI-lite
A synthetic oscillatory network
behaviour of the system, as not enough is known abut the molecular
! cI-lite
WithColE1 considerations in m
these
biology techniques to construct a
interactions inside the cell to make such a description realistic. Instead, hi
of transcriptional regulators
SyntheMc
Biology
PLtet01 repressilator and a compatible,
ENGINEERING
we hoped to identify possible classes of dynamic behaviour and taining the tet-repressible promo
Michael B. Elowitz & Stanislas Leibler intermediate stability variant of gf
determine which experimental parameters should be adjusted to obtain by IPTG interferes with repression
etters to nature sustained oscillations. and Physics, Princeton University, Princeton, pulse of IPTG might be capable o
Departments of Molecular Biology
New Jersey 08544, USA
Deterministic, continuous approximation repressilator-containing cells. A cu
b
.............................................................................................................................................. ing the two plasmids and grow
Three repressor-protein concentrations, pi, and theirdisplayed what appeared to be a s
Networks of interacting biomolecules steady state many essential carry out corresponding
Protein lifetime/mRNA lifetime, β
A
functions in living cells1, but the `design principles' underlying the ¯uorescence per cell after transfer
mRNA concentrations, mi (where i remain poorly under- were treated asindividual c
functioning of such intracellular networks
is lacI, tetR or cI) not shown). Because
stable
a Repressilator Reporter 30 8C. At least 100 individua
continuous dynamical variables. Each of these six molecular species
stood, despite intensive efforts including quantitative analysis of maintaining synchronization, we
2
participatestointhis systems . Here design and construction of a lating reactions.under they mi
relatively simple
transcription, translation complementary ¯uorescence intensity as the gr
we present a
and degradation single cells Here B
approach colonies were tracked manua
problem: the
we consider only to implement a particular function. We used three repressors are hund
synthetic network the symmetrical case in which all microcolonies consisting of
PLlac01
three quanti®ed.
identicaltranscriptional repressor systems that arenetwork, termed The kinetics of observation ti
except for their DNA-binding speci®cities. experiments, total the after not part of any
ampR natural biological clock3±5 to build an oscillating entering a stationary phase
tetR-lite system are determined by six coupled ®rst-order differential equations: ¯u
The timecourse of the
C
| | |
NATURE VOL 403 20 JANUARY 2000 www.nature.com
‡ Fig. 2. Temporal oscillation
dmi Ltet01
P a steady state
ˆ 2 mi ‡ a0 ! unstable
kanR dt n
…1 ‡ pj †
overallˆ increasecI in ¯uoresce
i lacI; tetR;
TetR TetR Maximum proteins cI; cell, α (! K )
pSC101 dpi gfp-aav 150 minutes, roughly threefo
j ˆ per lacI; tetR M
!PR ˆ 2 b…pi 2 mi †
origin dt time. The amplitude of oscil
! cI LacI where the number of protein copies levels produced from a given
GFP per cell of GFP ¯uorescence.
promoter type duringccontinuous growth is a0 in the presence of
exhibit oscillatory behaviou
saturating amounts of repressor (owing to the `leakiness' of the 6,000 6,000
lacI-lite 1 1
promoter), and a ‡ a0 in its absence; b denotes the ratio ofa Fourier a
determined by the protein
Proteins per cell
ColE1 4,000
0
4,000
0
! cI-lite decay rate to the mRNA decay rate;range of periods, as estimate
and n is a Hill coef®cient. Time is -1 0 500 1,000 -1 0 500 1,000
rescaled in units of the mRNA lifetime; protein concentrations are written
2,000 2,000
PLtet01 intervals, is 160 6 40 min (m
in units of KM, the number of repressors necessary to half-maximally
repress a promoter; and mRNA concentrations are rescaled by theirsiblin
GFP levels in the two 0
0 500 1,000
0
0 500 1000
Time (min) Time (min)
translation ef®ciency, the average number of proteins produced per perio
one another for long
mRNA molecule. The numerical solution of the model shown in Fig. 1c

17. Molecular Systems Biology (2006) doi:10.1038/msb4100090
& 2006 EMBO and Biology (2006) Group All rights reserved 1744-4292/06
Molecular Systems Nature Publishingdoi:10.1038/msb4100090
& 2006 EMBO and Nature Publishing Group All rights reserved 1744-4292/06
www.molecularsystemsbiology.com
www.molecularsystemsbiology.com
Article number: 45 Available online at www
Article number: 45
REVIEW
REVIEW
ENGINEERING
Towards synthesis of a minimal cell
Towards synthesis of a minimal cell
Anthony C Forster1,* and George M Church2,* step, yet detailed plans have not been published. Here,
1
Synthetic biology through reviewcell contain
Anthony C Forster1,* and George M Church2,*
Department of Pharmacology and Vanderbilt Institute of Chemical Biology,
biomo
step, yet detailed plans have not been published. Here,
attempt to outline the synthesis of a minimal
attempt to outline the synthesis of a minimal cell contain
the core cellular replication machinery, the pertin
Towards synthesis of a minimal cell
1
1 AC Forster and GM Church
the core cellular replication machinery, review the pertin
2
2
Vanderbilt University Medical
Kevin Channon , Elizabeth HC Broml
Department of Pharmacology Center, Nashville, TN, USA and
and Vanderbilt Institute of Chemical Biology,
Vanderbilt University Medical Center, Nashville, TN, USA and USA
Department of Genetics, Harvard Medical School, Boston, MA,
Department of Genetics,AC Forster, Department of Boston, MA, USA
Available literature and highlight gaps in knowledge that need filling
online at www.sciencedirect.com
literature and highlight gaps in knowledge that need filling
* Corresponding authors. Harvard Medical School, Pharmacology, Vanderbilt
* Corresponding authors. AC23rd Ave.S. at Pierce,of Pharmacology, Vanderbilt
University Medical Center, Forster, Department Nashville, TN 37232, USA.
University 615 936 Center, 23rdþ 1 615 at Pierce, Nashville, TN 37232, USA.
Tel.: þ 1 Medical 3112; Fax: Ave.S. 936 5555; Utility
E-mail: 1 615 936 3112; Fax:list 615very short, containing only Utility
Tel.: þ a.forster@vanderbilt.edu or is Synthetic biology through151 genes and 113 kbp. All engineering
þ 1 GM 936 5555;
Church, Department of Genetics, biomolecular design and
Harvard Medical School, 77 Avenue Louis Pasteur, NBRderived from rapidly growing field thatWoolfson1,2 knowledge
of theKevin Channon1Boston, is a E. coli and its a minimal cell has emerged in
E-mail: a.forster@vanderbilt.edu or GM Church, Department ofBoston,
genes are 238, ,Genetics,
Synthetic biology Synthesizing will advance
Elizabeth HC Bromley1 a minimal cell will advance knowledge
bacteriophages
Synthesizing and Derek N
Harvard Medical School, 77 Avenue Louis Pasteur, NBR 238, biological replication. Many hypotheses in replication and
MA 02115, USA. Tel.: þ 1 617 432 1278; Fax: þ 1 617 432 6513
(except global, multidisciplinary effort replication. biologists,such in replicationbiolo
a Fax: þ 1 hammerhead RNA subsystemsamong beForster in chemists, and
MA 02115, USA. Tel.: þ 1 617 432 1278;for the 617 432 6513 biologicalplant virus;
from a can only Many hypotheses a synthetic
tested
Symons, 1987), implyingthat has emerged individual allows a ‘synthetic’of(from atosynthetic biol
and
Received 7.5.06; accepted 26.7.06
subsystems can only be tested in such
Synthetic biology is a rapidly growing field thatproject. The meaning of wide variety
the in second, it subsystems studies Greek sunthesis
be grouped
engineers, physicists, and mathematicians. general approaches. We believe that
Received 7.5.06; accepted 26.7.06 into a small number of Broadly, the field
project. The meaning of ‘synthetic’ (from Greek sunthesis
a global, multidisciplinary effort among biologists, chemists,
will be compatible. In contrast Broadly, the field to lists together) discussed here bypasses the current reliance
put derived by be useful in defining and, hopefully, helping to
comparative
has two complementaryput together) discussed here bypasses the current reliance
engineers, physicists, and mathematicians. this will
has two complementary approaches, the biochemicallycells or understanding products:
goals: To improve macromolecular cell of
genomics or geneticgoals: To improve producesynthetic biology on exciting andlist area of synthetic biology.
understanding of develop the based broad
biological systems through mimicry and to
synthetic biology on cells or macromolecular cell products:
bio-
aim the to put together an organism from small molecules alo
is to put together an organism from small molecules alo
biological systems through mimicry and to produce bio-
does not contain with newgenes of unknown function or challen- the breadth of topics that
orthogonal systems any functions. Here we review is aim For this review, because of
Construction of a chemical system capable reference to the concept of syntheticsimplest approach emerging field, we found it necessary to may
The simplest approach for creating an artificial cell
ging membrane is, of replication is close to a contributeunderstood, pastan artificial cell of R
capable of replication to this for creating
area specifically with
Construction of a chemical systemspace, thatproteins;ofso itwith new functions.studies fromwe review the may
orthogonala systems The
components for, evolvingfullyrefer to classic Here the
and evolution, fed only by small molecule nutrients, is now functionaland
biology
evolution, fed only by small molecule nutrients, is now hierarchy by evolving an RNA polymerase made decades, reviews of R
an RNA polymerasefive years, exclusively
two
and replicating ‘platform’ (Szostak etin few known the past made exclusively
approaches to synthetic by
accurately bytest,generating new apply our and for life. The al, 2001) to replace all protein components
various areas from gaps as well as more
conceivable. This could be achieved byspecifically with referenceet al, work conceptthree years.
area stepwise integration
conceivable. This could be achieved to stepwise and
systems advance,
integration (Szostak to the from the past of synthetic
understanding of recent 2001) to replace all protein components
constitute only about seven issue inall of replicating and evolving systems
with this genes, vitro which are predicted
of decades of work on the reconstitution of In keepingRNA and of Currentvitro replicating and evolving systems (e.g. to replace
biological systems.
of decades of work on the reconstitution of DNA, wethat is,on thehierarchy of components for, (e.g. to replace
biology of biomolecule-based (Table indesign Mills et al, column).
Opinion in Structural DNA, RNA and
protein syntheses from to becomponents. Such a minimal areplicase;in the left 1967). But in comparison with a purifi
RNA space, and
Biology, focus largely
protein pure forengineeringmodification componentsreplicase; Mills et al, 1967). But in comparison with a purifi
syntheses from pure components. Such a minimal
and I, bold Synthetic biology space: hierarchies of components,
and
protein-based system,ofapproaches guaranteed to arrive soo
interactions and is neither
the viewpoint to structural biology, syntheticit is neither aguaranteed
approaches of generating new courtesy the hierarchy is set of basic units—amino it and functional to arrive soo
cell project would initially define the components sufficient
cell project would initially define the components sufficient
From systems. protein-based system, recent
each subsystem, allow detailed kineticribosome and
Addresses
kinetic analyses Bristol nor tell us more. base of
At the A protein-based system
for each subsystem, allow detailed Chemistry, University of Bristol, and BS8 1TS, UKdetermination (Diaconuand lipidswill connect with, a
for breakthroughs intoanalysesadvance, and apply protein-based system 1will connect with, a
systems synthesisBristol, Bristol reveal UKus more.nucleicexisting biological systems. One lik
School of 1 nor tell acids, A
test, of structure 1TD, more about, existing biological systems. Life, lik
our understanding of Life,
acids, sugars (Figure 1).
lead to improved in vitro methods Biochemistry,
vitro methods for synthesis of
lead to improved in et al, 2005; Ogle and University
Department of for
2
of reveal more about,
BS8 level of complexity above these are what might be termed
Ramakrishnan, 2005),tectons. be understood from supramolecular chem-
significant three-
biological systems. In machine, cannotThis termitisisborrowedsimply by studying it and
biopolymers, therapeutics and biosensors. Completion keeping with this issue to simply programmed mol- and
biopolymers, therapeutics and biosensors. Completion
Corresponding author: Woolfson, Derek N (D.N.Woolfson@bristol.ac.uk) cannot be understood
machine, of Current by studying it
dimensional information is lacking for istry [1], be put together describeits parts. Along the w
would yield a functionally and structurally understood
yield a functionally and structurally understood only where ofused from its parts. Along the w
parts; it must also 3% and nanoscale building blocks [2]. An
parts; it must also be put together from
ecular components
the
would
products: aOpinion in Structural Biology, we focusnucleic acid tecton would bedesign
Opinion Structural Biology life to synthesizing a cell, we might discover new biochemi
Current few
for synthetic life to example a largely on the
self-replicating biosystem. Safety concernsin RNA modification proteins and cell, we the information for further assem-
self-replicating biosystem. Safety concerns for synthetic2008, 18:491–498 synthesizing a ofaminoacyl- discover anew oligo- might short biochem
and engineering of
will be alleviated by extreme dependence on elaborate
This review comes from on elaborate
dependence
will be alleviated by extreme Engineering and design(Table I, right column). essential for replication,
tRNA synthetases essential containing with unsuspected macromo
nucleotide
a themed issue on biomolecule-based components and
functions While some of the other tectons. Similarly, an
for replication, unsuspected macromo
functions bly through interactions
laboratory reagents and conditionsDekfor viability.Regan to be solved at high based tecton would beunrecognized patterns
laboratory reagents and conditions for viability. Our
states systems. and Lynne Our
and by Woolfson remain
Edited
complexes
cular modifications or previously unrecognized patterns
cular modificationsresolution,
amino acid or previously a polypeptide designed
proposed minimal genome is 113 kbp long and contains
proposed minimal genome is 113 kbp long and contains coordinated expression. of self-assembling a-helix or b-strands.
to form stretches
coordinated expression.
a draft Available online 5th August 2008 structure for any replicating system an artificial, aprotein-bas
three-dimensional Importantly, a tecton is something more than simple
How good a model would an artificial, that the
Addresses matter the history of biology. a model would
151 genes. We detail building blocks already in place and
151 genes. We detail building blocks already in place and
0959-440X/$ – see front
is a major milestone in
How good element of secondary structure: It implies protein-bas
Ltd.
completion. Chemistry, minimal cell be for natural cells? The only cellular alternat
major hurdles to overcome for 1# 2008 Elsevier of All rights reserved. University of element contains informationstructures.further assembly
about only cellular alternat
its
School is a perturbed natural cell, an BS8 1TS,Combiningsystem ev
Bristol, Bristol incredibly complex system e
into prescribed higher order UK
complex
tec-
Molecular pathwaysBiology 22 August 2006;
Figure 1 A minimal cell containing biological macromolecules and Systems 2DOI 10.1016/j.sbi.2008.06.006 tons leads to the next level in the hierarchy, in which self-
proposed to be necessary and sufficient for replication from small molecule
doi:10.1038/msb4100090 Department of Biochemistry,the simplest ofof Bristol, Bristol BS8 1TD,pro- ba
for University cells. are much simpler purified system bainteractions UK
assembled units A formed through purified system

18. REVERSE
SYSTEMS
ENGINEERING
ENGINEERING
BIOLOGY
DEFINE
DEFINE
MATH
HYPOTHESIS
ELEMENTS
INTERACTION
DESCRIPTION

19. REVERSE
ENGINEERING
Ideally:
all
the
elements
of
a
system…
genes
isoforms
Proteins
(with
all
modificaMons)
Metabolites
DEFINE
all
together…
ELEMENTS
FOCUS
THE
SELECT
BASED
ON
QUESTION
THAT
Genes
being
modified
in
T-­‐cell
acMvaMon?
How
do
they
interact?

20. What
is
Piece
1:
The
Reverse
connecMons
Engineering?

21. REVERSE
ENGINEERING
DEFINE
ELEMENTS

22. REVERSE
SYSTEMS
ENGINEERING
BIOLOGY
DEFINE
DEFINE
ELEMENTS
INTERACTION
PIECE
1

23. A
B
A
B
A
B
1
KNOWLEDGE
2
DATA

24. INTERACTION
IDENTIFICATION
BY
DATA
Wisdom of crowds for robust gene network inference
Daniel Marbach1,2,11, James C Costello3–5,11, Robert Küffner6,11, Nicole M Vega3–5, Robert J Prill7,
Diogo M Camacho3–5,10, Kyle R Allison3–5, The DREAM5 Consortium8, Manolis Kellis1,2, James J Collins3–5,9 &
Gustavo Stolovitzky7
Reconstructing gene regulatory networks from high-throughput successfully used to address many biological problems8–11, yet
data is a long-standing challenge. Through the Dialogue on when applied to the same data, they can generate disparate sets
Reverse Engineering Assessment and Methods (DREAM) project, of predicted interactions2,3.
we performed a comprehensive blind assessment of over 30 Understanding the advantages and limitations of different network
© 2012 Nature America, Inc. All rights reserved.
network inference methods on Escherichia coli, Staphylococcus inference methods is critical for their effective application in a given
aureus, Saccharomyces cerevisiae and in silico microarray data. We biological context. The DREAM project is a framework to enable
characterize the performance, data requirements and inherent such an assessment through standardized performance metrics and
biases of different inference approaches, and we provide guidelines common benchmarks12 (http://www.the-dream-project.org/).
for algorithm application and development. We observed that no DREAM is organized around annual challenges, whereby the com-
single inference method performs optimally across all data sets. munity of network inference experts is solicited to run their algo-
h_p://www.the-­‐dream-­‐project.org/
In contrast, integration of predictions from multiple inference rithms on benchmark data sets, participating teams submit their
methods shows robust and high performance across diverse data solutions to the challenge and the submissions are evaluated12–14.
sets. We thereby constructed high-confidence networks for E. coli Here we present the results for the transcriptional network
and S. aureus, each comprising ~1,700 transcriptional interactions inference challenge from DREAM5, the fifth annual set of DREAM
at a precision of ~50%. We experimentally tested 53 previously systems biology challenges. The community of network infer-
unobserved regulatory interactions in E. coli, of which 23 (43%) ence experts was invited to infer genome-scale transcriptional
were supported. Our results establish community-based methods as regulatory networks from gene-expression microarray data sets
a powerful and robust tool for the inference of transcriptional gene for a prokaryotic model organism (E. coli), a eukaryotic model
regulatory networks. organism (S. cerevisiae), a human pathogen (S. aureus) and an
in silico benchmark (Fig. 1).
‘The wisdom of crowds’ refers to the phenomenon in which the The predictions made from this challenge enabled the com-
collective knowledge of a community is greater than the knowl- prehensive characterization of network inference methods
edge of any individual1. Based on this concept, we developed across different species and data sets, providing insights into
a community approach to address one of the long-standing method performance, data requirements and inherent biases.
challenges in molecular and computational biology, which is to We found that the performance of inference methods varies,
uncover and model gene regulatory networks. Genome-scale with a different method performing best in each setting. Taking
inference of transcriptional gene regulation has become pos- advantage of variation, we integrated predictions across infer-
sible with the advent of high-throughput technologies such as ence methods and demonstrated that the resulting community-
microarrays and RNA sequencing, as they provide snapshots of based consensus networks are robust across species and data
the transcriptome under many tested experimental conditions. sets, achieving the best overall performance by far. Finally, we

25. INTERACTION
IDENTIFICATION
BY
DATA
A.
CORRELATION
A
B
A
B
C
D
C
D

26. Segal
et
al.
2004
Nature
GeneMcs.

27. Novershtern
et
al.
2008
Am
J
Respir
Cell
Mol
Biol

28. Stuart
et
al.
2003
Science

29. INTERACTION
IDENTIFICATION
BY
DATA
B.
MUTUAL
INFORMATION
MI(A,B)
measures
their
mutual
dependence
A
B
A
B
C
D
C
D