Upload
others
View
9
Download
0
Embed Size (px)
Citation preview
Programming DNA sequences to engineer multi-component chemical systems
Neil Dalchau
Computational Science Laboratory
Biological Computation Group
Synthetic Biology and Control WorkshopWorcester College, University of Oxford
10 - 12th September 2014
2050: Doctor in a cell (Udi Shapiro)
Programmable molecular computer
computational
science
laboratory
Programming biological systems
Molecules Cells Colonies
Computer Aided Design software
computational
science
laboratory
Computational DNA Drugs
• Perform logical computation before releasing drug
• Uses restriction enzymes
Simplified (omitting the “no” pathway)
An automaton sequentially reading the string PPAP2B, GSTP1, PIM1, HPS (known cancer indicators) and sequentially cutting the DNA hairpin until a ssDNA drug (Vitravene) is released.
Vitravene (GCGTTTGCTCTTCTTCTTGCG)
(Restriction Enzymes)
Building stuff with DNA…
computational
science
laboratory
Using DNA to implement algorithms
DNA Strand Displacement
Microsoft Research Outreach
computational
science
laboratory
VIDEO
DNA structure abstraction
8
computational
science
laboratory
DNA domains
Short complementary domains bind reversibly
Long complementary domains bind irreversibly
9
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
10
Input 1 Input 2
Output
Substrate
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
11
Input 2
Output
Substrate
Input 1
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
12
Input 2
Output
Substrate
Input 1
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
13
Input 2
Output
Substrate
Input 1
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
14
Output
Substrate
Input 1
Input 2
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
15
Substrate
Input 1 Input 2
Output
computational
science
laboratory
Strand displacement logic circuit
Output = Input1 AND Input2
16
Substrate
Input 1 Input 2
Output
computational
science
laboratory
Domain abstraction
17
tb --> (5') TACCAA (3')tx --> (5') TATTCC (3')to --> (5') GTCA (3')b --> (5') CCCTTTTCTAAACTAAACAA (3')x --> (5') CCCAAAACAAAACAAAACAA (3')
computational
science
laboratory
DNA strand displacement (DSD) language
Programming DNA circuits
Step 1: Program circuit design Step 2: Compile circuit behaviour Step 3: Simulate circuit
computational
science
laboratory
Approximate Majority algorithm
X + Y 2BX + B 2XY + B 2Y
Y (Minority)
X (Majority)
X (Majority)
computational
science
laboratory
Autocatalytic component
X + B 2X + PX
BX
N.B. Two-Domain DNA Strand Displacement was proposed by Luca Cardelli (DCM 2010)
X XPX
High-level specification
DNA-level implementation
computational
science
laboratory
Join gate Fork gate
Autocatalytic component
X + B 2X + PXHigh-level specification
DNA-level implementation
computational
science
laboratory
Component characterisation
X + B 2X + PX
computational
science
laboratory
Data - - - Model fitting+ + Model fitting
Component characterisation
X + B 2X + PX
Data - - - Model fitting+ + Model fitting
computational
science
laboratory
X + Y 2B + PB
Y + B 2Y + PY
Predicting dynamics of the full circuit
X + Y 2B X + B 2X Y + B 2Y
Experimental Data - - - Model prediction
computational
science
laboratory
Predicting dynamics of the full circuit
X + Y 2B X + B 2X Y + B 2Y
Experimental Data - - - Model prediction
computational
science
laboratory
Speeding up with higher concentrations
X + Y 2B X + B 2X Y + B 2Y
Experimental Data - - - Model prediction
computational
science
laboratory
Consensus over the initial majority
computational
science
laboratory
Summary
• DNA is an excellent programmable material
• CRNs offer a rich set of behaviours that can be recapitulated using DNA-based strategies
• Detailed molecular models can be programmed using Visual DSD
– Bayesian parameter inference enables characterisation of the detailed models
– Predictions can be very good
computational
science
laboratory
Consensus in space
Dalchau, Seelig, Phillips (DNA 2014)
computational
science
laboratory
Conclusions
• Manipulating DNA sequences enables the creation of complex circuits
– Implement algorithms using DNA strand displacement
– Fine-tune and sometimes completely change sensitivity to input signals
• Detailed mathematical modelling, in combination with parameter inference, enables quantification of specific biochemical processes
– Where are the bottlenecks? Where is there interference?
– How do we optimally modify the system to improve performance?
Acknowledgements
Yuan-Jyue ChenGeorg Seelig
David Soloveichik
Andrew PhillipsBoyan YordanovLuca CardelliStephen EmmottMatt LakinFilippo PoloColin GravillRasmus Petersen
University of Washington
UCSF
Microsoft Research Cambridge
Paul GrantJames BrownJim HaseloffJim Ajioka
University of Cambridge
computational
science
laboratory
http://research.microsoft.com/science/tools