1

Design Automation for DNA Self-Assembled Nanostructures

Constantin Pistol, Alvin R. Lebeck, Christopher Dwyer

Duke University

Design Automation Conference - July 27th, 2006

3

DNA Nano-Assembly Today

100 nm

How to drive this forward ?

Sung Ha Park, Constantin Pistol, Sang Jung Ahn, John H. Reif, Alvin R. Lebeck, Chris Dwyer, Thomas H. LaBean - "Finite-size, Fully-Addressable DNA Tile LatticesForme

by Hierarchical Assembly Procedures", Angewandte Chemie No.5/2006 Y. He, Y. Tian, Y. Chen, Z. Deng, A. E. Ribbe and C. Mao, Sequence Symmetry as a Tool for Designing DNA Nanostructures, Angewandte Chemie International Edition, 44 (2005), pp. 6694-6696.

J. Sharma, R. Chhabra, Y. Liu, Y. Ke and H. Yan, DNA-Templated Self-Assembly of Two-Dimensional and Periodical Gold Nanoparticle Arrays, Angewandte Chemie International Edition, 45 (2006), pp. 730-735.

Constantin Pistol, Alvin R. Lebeck, Dan Sorin, Chris Dwyer - “Nanoscale Device Integration on DNA Self-Asembled Nanostructures”, Duke University, 2006

M. Mertig, W. Pompe: Biomimetic fabrication of DNA-based metallic nanowires and networks, in: Nanobiotechnology - Concepts, Applications and Perspectives, Ed. by C.M. Niemeyer, C.A. Mirkin, WILEY-VCH Verlag GmbH & Co. KgaA, Weinheim, 2004, p.256-277

P. W. K. Rothemund, Folding DNA to create nanoscale shapes and patterns, Nature, 440 (2006), pp. 297-302.

Nanowire Transistors, Gate Electrodes, and Their Directed Self-Assembly K. Skinner, R. L. Carroll, S. Washburn, and C. Dwyer. The 72nd Southeastern Section of the American Physical Society (SESAPS), November 2005

4

Goal

• Automate the design of DNA-based self-assembled structures

- Find optimized sequences for given target

structures

- Define metrics for quantitative design comparison

- Apply self-assembly specific design rules

- Expose useful trade-offs to designers

INPUT Structure Motif Seq. space

CONSTRAINTS Set optimization Design rules Trade-offs

OUTPUT Seq. map files manufacture ready

5

Outline

1. DNA Basics

2. DNA Self-Assembly and DNA Motifs

3. Metrics and Design Rules

4. Implementation

5. Evaluation

6. Conclusion

6

DNA Basics

• A DNA strand is:

- A linear array of bases (A, T, G, and C)

- Directional (one end is distinct from the other)

- In nature, the source of genetic information

• DNA will form a double helix:

- When the bases on each strand (aligned head-to-

toe) are complementary: A with T, and G with C

- But only under “natural” environmental conditions

such as (low) temperatures (sequence dependent) and in

an ionic solution.

A

G

C

T

T

A

G

A

T

C

A

T

C

G

A

A

T

C

T

A

G

T

How does it form?

7

DNA Basics

• DNA hybridization is the process that forms the double helix

• Random diffusion: power of self-assembly

• Sequence and temperature controls the hybridization event

• Reverse process called melting

- Melting temperature (Tm) is sequence dependent

- G-C pairs ~2x as strong as A-T pairs

T

8

DNA Basics

• A common form of the double helix has some well-known

geometric properties:

- 3.4 Å per base pitch along the helix

- One complete turn between every 10th and 11th base

• Flexibility: the bonds along the sugar-phosphodiester

backbone of each strand can rotate

- Single stranded DNA has ~5 - 10 nm persistence length

- Double stranded DNA has ~50nm persistence length

Persistence length is “straight length”

9

Outline

1. DNA Basics

2. DNA Self-Assembly and DNA Motifs

3. Metrics and Design Rules

4. Implementation

5. Evaluation

6. Conclusion

10

Self-Assembly

• Self-assembly is ubiquitous in nature

• Generally defined as spontaneously generated order

• Thermodynamics drive the self-assembly process

- We can guide the process by the choice of

materials and environmental conditions

A

B

TA·B

Advantages ?

11

Nanoscale self-assembly advantages

• Size – feature sizes in the 1 - 20nm range

• Count – simple parallel assembly (1014 – 1016)

• Infrastructure – $460 billion chemical manufacturing industry

• New material properties – leverage quantum effects

• Potential for dense nano-scale active devices

Why use DNA ?

* 2002 US Census

12

• DNA can provide substrate for fabricating nano-devices

Why DNA Based Self Assembly

• Nano-scale components placed / interconnected on substrate

- Crossed carbon nanotube FETs

- Ring-gated FETs

- Nanowires

- Quantum dots

- Precise binding rules

- Nanometer pitch

• Challenge: specify DNA sequences for

- Intended geometry

- Thermodynamic stability

How to address this challenge?

13

Motifs and Hierarchical Assembly

• Complex designs built from small set of building blocks (motifs)

• Many possible motifs

- Junctions (to form triangles, corners, etc.)

- Sticky-ends: single strand of DNA protruding out of a helix

• Two motifs with complementary sticky-ends bind to form a composite motif

• Composite motifs can bind with other composite motifs to form larger composite motifs

Structure design = hierarchical structuring of motifs

T3.4 ÅT

Sticky ends = “smart” glue

14

Cruciform Motif

• Motif* has 9 DNA single strands

- Arms end in 5bp sticky ends

• Not exactly flat*

- Some structural curvature

• Core strand can be functionalized

Shell 1

Shell 2Shell 3

Shell 4

Arm 1

Arm 2

Arm 3

Arm 4 Core

* H. Yan, S. H. Park, L. Feng, G. Finkelstein, J. H. Reif, and T. H. LaBean, "4x4 DNA Tile and Lattices: Characterization, Self-Assembly, and Metallization of a Novel DNA Nanostructure Motif," in Proceedings of the Ninth International Meeting on DNA Based Computers (DNA9), 2003.

- Molecular addressability

- Two sticky ends per arm

- Attach nanoscale components

15

Target System – 4x4 Grid

• 16 active access points on flat 60nm x 60nm grid

• “Nano-board” scaffold – “pin” device elements to it

T16

T2

T1

• Hierarchical 2-level assembly

60nm

Sung Ha Park, Constantin Pistol, Sang Jung Ahn, John H. Reif, Alvin R. Lebeck, Chris Dwyer, Thomas H. LaBean - "Finite-size, Fully-Addressable DNA Tile Lattices Formed by Hierarchical Assembly Procedures", Angewandte Chemie No. 5/2006

How to design this system?Core and shells reused - fixed sequences

16

Design Automation – 4x4 Grid

• Complex simultaneous interaction – 16 motifs with 48 arms

• Design targets:

Target 1 : design optimized 96 sticky end set

Target 2 : eliminate curvature (flat grid)

Target 3 : control & optimize self-assembly process

DNA Design Automation Software

T1 T2 T3

INPUT Structure Motif Seq. space

Constraints OUTPUT

Sequence map

How to evaluate these targets? What metrics?

17

Outline

1. DNA Basics

2. DNA Self-Assembly and DNA Motifs

3. Metrics and Design Rules

4. Implementation

5. Evaluation

6. Conclusion

18

Target 1: Sequence design

• Generate optimized DNA sequence set for a given target structure

• Two metrics

- SEM: average single-interaction energy measure (stability)

- TLM: average target-interaction likelihood measure (accuracy)

• Find sequences that

- Minimize strength of unintentional interactions

- Maximize strength of intentional interactions

• Designer can add tradeoff between accuracy and stability

19

SEM and TLM Metrics

Specific Tm

No

n-S

pec

ific

Tm

•

1:1

diagonal

SEM = Spec[strand]

TLMSEM and TLM for single sequence point

Specific Tm : Melting temp. with the complement strand

Non-Specific Tm : Highest melting temp. with a non-complementary strand

* N. Le Novere, "MELTING, computing the melting temperature of nucleic acid duplex," Bioinformatics, vol. 17, pp. 1226-1227, 2001

Melting temp. of strand pairs calculated using modified version of MELTING4* software

NonSpec[strand]

20

SEM and TLM Metrics – Stability and Accuracy

- Set E + C1 has high SEM and low TLM

G-C-CC-G-G

A-C-CT-G-G

A-T-AT-A-T

EXISTING (E) CANDIDATE 1 (C1) CANDIDATE 2 (C2)

- High stability (melting temperature) but low accuracy (more defects)

• Assume sequences E are present in system (sticky end)

• Need to add an additional sticky end

• Which is better: C1 or C2 ?

- Set E + C2 has lower SEM and high TLM

- Lower stability but higher accuracy (less defects)

G-C pair stronger than A-T

21

Target 2: Flat Grid

- Corrugation design rule – alternate motif normals

Face UP

Face DOWN

22

Target 3: Optimize Assembly

- Thermal ordering design rules - hierarchy of sub-products

1st

2nd

3rd

4th

Thermal groups

Temperature

High

Low

23

Outline

1. DNA Basics

2. DNA Self-Assembly and DNA Motifs

3. Metrics and Design Rules

4. Implementation

5. Evaluation

6. Conclusion

24

• Exhaustive thermodynamic search

Thermodynamic Optimization Software

• Optimize target design against TLM and SEM metrics given:

• Evaluate each sticky-end sequence

- Target topology

- Basic motif design

- Against all other candidate and motif sequences

- Map their mutual interaction

• High computational cost

- 4x4 grid: 6 CPU-years to design and verify

- Parallel implementation

- Creates detailed sequence interaction

map

Str

an

d A

Strand B

Maximum Tm(A,B)

Strand Function1 – 4 : Shell5 : Core6 – 9 : Arms

°C

25

• Modified nearest-neighbor algorithm using MELTING4* tool

Thermodynamic Interaction

• The number of interactions to evaluate depends on:

- Length of sticky-ends

- Extent of fixed motif sequences

- Energy contribution of each base also depends on neighbors

- Handles internal and terminal mismatches

Sticky-end length Sticky-end Interactions

5bp > 500,000

10bp > 500,000,000,000

• Calculate melting temperature for each interacting pair – used for SEM, TLM metrics

* N. Le Novere, "MELTING, computing the melting temperature of nucleic acid duplex," Bioinformatics, vol. 17, pp. 1226-1227, 2001

26

Design Automation Suite (DAS)

• Input: structure + motifs + sequence space

• Software runs thermodynamic analysis

- Generates metric-annotated DNA sequence sets

• Designer can customize set

- Default: optimize for maximal TLM

- SF (Stability Factor): trade TLM for SEM

• Design rules are applied on candidate set

- Corrugation and thermal ordering

• Output: sequence map files

- For order and manufacturing

Co

ns

traints

27

Outline

1. DNA Basics

2. DNA Self-Assembly and DNA Motifs

3. Metrics and Design Rules

4. Implementation

5. Evaluation

6. Conclusion

• Target systems and methods• Theoretical metric-based evaluation• Experimental evaluation

28

Evaluation: target systems

• Large design - 96 sticky-ends set

- 4x4 grid with 16 motifs

• Small design – 20 sticky-ends set

- AB Polymer* with 4 motifs

* H. Yan, S. H. Park, L. Feng, G. Finkelstein, J. H. Reif, and T. H. LaBean, "4x4 DNA Tile and Lattices: Characterization, Self-Assembly, and Metallization of a Novel DNA Nanostructure Motif," in Proceedings of the Ninth International Meeting on DNA Based Computers (DNA9), 2003.

A,B: Two sets of fixed sequences

What is the impact of fixed sequences ?

A-Core B-Core

29

Impact of fixed sequences

- Accuracy in arm space with different fixed sequence sets

A-only, B-only:

- same target structure as AB Cores

- use a single core/shells set (fewer fixed sequences)

Accuracy of the 5 bp arm space

0

2

4

6

8

10

12

1 28 55 82 109 136 163 190 217 244 271 298 325 352 379 406 433 460 487 514 541 568 595

Arm Identifier

Ac

cu

rac

y (

TL

M)

AB Cores

A-only

B-only

No Core

More fixed sequences = fewer “good” arms to choose from

30

• Random sequence sets

Evaluation: Sequence Design Methods

• DNA Design Automation Suite (DAS)

+ Thermodynamic analysis of all possible interactions

+ High flexibility, detailed sequence interaction map

- Computationally expensive

• Original 20-Arm design – based on existing sequence design tools*

- Text-distance primitive

- GC content as energy approximation

+ Fast but no result / local minimum for large seq. spaces

- No guarantees of optimality

*N. C. Seeman, "De Novo Design of Sequences for Nucleic Acid Structural Engineering," Biomolecular Structure & Dynamics, vol. 8, pp. 573-581, 1990.

+ Fast

31

SEM and TLM (20 Arm)

0

2

4

6

8

10

12

14

16

18

20

Upper Bound DAS - B CoreSF=4

DAS - AB SplitSF=7

Original AB Random

SEM

TLM

Sequence design evaluation – 20-Arm

• 20-Arm target system: upper bounds and original design

DAS designs significantly improve both accuracy and stability

32

Sequence design evaluation – 96-Arm

• 96-Arm target system: B-only, AB Core and Random

-20

-15

-10

-5

0

5

10

15

20

25

0 5 10 15 20 25

Specific Tm

No

n-S

pec

ific

Tm

RandomAB CoresB-Only

1:1 diagonal

B-only = highest accuracy (TLM)

SEM and TLM (96 Arm)

0.00

2.00

4.00

6.00

8.00

10.00

12.00

B-Only AB Cores Random

SEM

TLM

33

Experimental results – 96-Arm target system

• Atomic Force Microscopy – 4x4 grids

60nm

• Molecular height map on flat plane

Sung Ha Park, Constantin Pistol, Sang Jung Ahn, John H. Reif, Alvin R. Lebeck, Chris Dwyer, Thomas H. LaBean - "Finite-size, Fully-Addressable DNA Tile Lattices Formed by Hierarchical Assembly Procedures", Angewandte Chemie No. 5/2006

34

Grid – no detectable curvature

• AFM height section shows flat grid

• Corrugation design rules successful

35

DNA scaffold – experimental validation

60nm

Manufacturing scale: ~1014 letters/mL!

Trivia: The collection of books and manuscripts in the Library of Congress contains ~1014 letters.

36

Conclusion

• Design automation for DNA-based self-assembled structures

• Step forward towards nano-scale device engineering

• Input: structure, motifs, sequence space

• Constraints:

- Optimized sequences for given target structures

- Self-assembly specific design rules

- Design trade-offs and metrics for quantitative design

comparison

• Output: sequence map for manufacturing

37

Thank you!

Design Automation Conference - July 27th, 2006

38

0

2

4

6

8

10

12

14

16

0 0.5 1 2 3 4 5 6 7SF

TL

M

AB TLM

B_Only TLM

0

2

4

6

8

10

12

14

16

0 0.5 1 2 3 4 5 6 7SF

SE

M

AB SEM

B_Only SEM

Balanced designs

• Trade-off: accuracy for stability (TLM for SEM)

SF: Stability Factor – increases SEM, decreases TLM

SF Impact on SEM and TLM (20 Arms)

39

SEM and TLM: Set Metrics

• Average of single sequence metrics

Tm[seqi,seqj] = melting temperature of seqi and seqj

comp[seqi] = sequence complement of seqi