In-vitro Molecular Computing Based on DNA Strandshinze/hinze-slides... · 2015-09-15 · In-vitro Molecular Computing Based on DNA Strands An unconventional computing concept inspired

Motivation Operations on DNA Solution to Knapsack Problem Perspectives

In-vitro Molecular ComputingBased on DNA Strands

An unconventional computing concept inspired by nature

PD Dr. Thomas Hinze

Brandenburg University of Technology Cottbus – SenftenbergInstitute of Computer Science and Information and Media Technology

[email protected]

In-vitro Molecular Computing Based on DNA Strands Thomas Hinze


DNA as an Excellent Data Storage Medium

high storagedensity up to

1 bit / nm3

L.M. Adleman. Molecular Computation of Solutions to Combinatorial Problems. Science 266:1021-1024, 1994




A C GT


1

base pair in strand ofdeoxyribonucleic acid (DNA)

00 01 10 11

2 bit per nucleotide or per

bit / nm3





A

A

A

A

C

A

C

C

T

G

T

T

T

G

T

G

A C GT


1


00 01 10 11


5’

5’ 3’

3’

2 nm3 / base pair

bit / nm3





A

A

A

A

C

A

C

C

T

G

T

T

T

G

T

G

A C GT


1


00 01 10 11


electronicflash memory card

30.001 bit / nm

corresponds to

128 Gbit / 146 qmm

5’

5’ 3’

3’

2 nm3 / base pair

bit / nm3





high storagepersistence up toseveral thousand years

W. Miller et al. Sequencing the nuclear genome of the extinct woolly mammoth. Nature 456:387-391, 2008




under optimalenvironmental conditions






www.zmescience.com

from >20,000 years old mammothapprox. 80% of genome reconstructed







www.zmescience.com

from >20,000 years old mammothapprox. 80% of genome reconstructed


floppy disk: hard disk: DVD (expected):5 ... 10 years 10 ... 15 years 30 ... 50 years






highly efficientchemical processingby low energy consumption

L. Kari. Arrival of biological mathematics. The Mathematical Intelligencer 19(2):9-22, 1997




: hydrogen bond: covalent bond

Guanine

Cytosine

3’

3’

5’

Thymine

5’

Adenine

A

T

G

C

A

T

G

C

A G

T C

(break or form

internucleotide

chemical bond)

per Joule

elementary operations

up to 1018


O

N

C

C

C

N

NN

CH

O

C

H

H

N

H

N

C

C

N

NN

CH

CH

HH

N

C

O

N

H

C

N

C

C

C

H

N

CH3

H

N

C

O

N

CC

C HO

CH2

CH

CH CH2

CH

2’3’

1’4’

5’

O

CH2

CH

CH CH2

CH

2’3’

1’4’

5’

OH

O

P OO

O

OH

OH

O

CH2

CHCH

CH2

CH

4’

5’

1’

3’2’

CH2

CHCH

CH2

CH

4’

5’

1’

3’2’

OH

P OOO

O

O





: hydrogen bond: covalent bond

Guanine

Cytosine

3’

3’

5’

Thymine

5’

Adenine

A

T

G

C

A

T

G

C

A G

T C

(break or form

internucleotide

chemical bond)

per Joule

elementary operations

up to 1018

10160 1

bit set

or reset

electronically


O

N

C

C

C

N

NN

CH

O

C

H

H

N

H

N

C

C

N

NN

CH

CH

HH

N

C

O

N

H

C

N

C

C

C

H

N

CH3

H

N

C

O

N

CC

C HO

CH2

CH

CH CH2

CH

2’3’

1’4’

5’

O

CH2

CH

CH CH2

CH

2’3’

1’4’

5’

OH

O

P OO

O

OH

OH

O

CH2

CHCH

CH2

CH

4’

5’

1’

3’2’

CH2

CHCH

CH2

CH

4’

5’

1’

3’2’

OH

P OOO

O

O





massivedataparallelism





up to 1015

molecules

DNA and other

massivedataparallelism

DNA strands

per test tube (2ml)

simultaneous

and autonomous

molecular interactions

without central control




on DNA

Operations

2.



Operations on DNA (Selection)Gaining DNA strands

• Synthesis (oligos) 5’−ACGGAAC−3’ A C G G A A C

• Isolation (like plasmids or genomic DNA from organisms)

Handling DNA solutions

• Union (merge)

• Split (aliquot)

Forming and breaking hydrogen bonds

• Annealing (hybridisation)T T G C C T T

A C G G A A C

T

A

T

C

G

G

C

G

C

A

T

A

T

C

• Melting (denaturation) T

A

T

C

G

G

C

G

C

A

T

A

T

C

T T G C C T T

A C G G A A C







• Union (merge)

• Split (aliquot)



A C G G A A C

T

A

T

C

G

G

C

G

C

A

T

A

T

C


A

T

C

G

G

C

G

C

A

T

A

T

C

T T G C C T T

A C G G A A C







• Union (merge)

• Split (aliquot)



A C G G A A C

T

A

T

C

G

G

C

G

C

A

T

A

T

C


A

T

C

G

G

C

G

C

A

T

A

T

C

T T G C C T T

A C G G A A C



Operations on DNA (Selection)Enzymatically catalysed reactions

• Ligation (concatenation)................T

A

C

G

G

C A P

T A

T

T

A

P T

A

C

G

G

C

T

A

A

T

T

A

• Digestion (cleavage).....................A

T

G

C

C

G

G

C

C

G

A

THinP1 I

A

T

G

C G C P

C G C

G

A

T

PG

C

C

G

G

C

C

G

5’

3’

3’

5’

• Labelling (strand end modification)A

T

G

C

C

G

G

C

C

G

A

T

A

T

G

C

C

G

G

C

C

G

A

T

P

P+ 5’−Phosphat

• Polymerisation (completion)..........A G C

G

G

C

C A A

T

G

C

C

G

G

C

• PCR (polymerase chain reaction)... ........ duplicate strands

Separation and analysis of DNA strands

• Affinity purification (sep. by biotin)..A C

G

C

G

T

A

A

T

G

C TC

B T

A

A

T

T

A + 5’−Biotin TC

B T

A

A

T

T

A

• Gel electrophoresis (sep. by length) sort and detect strands

• Sequencing (readout).......................A C G G A A C 5’−ACGGAAC−3’



Operations on DNA (Selection)Enzymatically catalysed reactions

• Ligation (concatenation)................T

A

C

G

G

C A P

T A

T

T

A

P T

A

C

G

G

C

T

A

A

T

T

A

• Digestion (cleavage).....................A

T

G

C

C

G

G

C

C

G

A

THinP1 I

A

T

G

C G C P

C G C

G

A

T

PG

C

C

G

G

C

C

G

5’

3’

3’

5’

• Labelling (strand end modification)A

T

G

C

C

G

G

C

C

G

A

T

A

T

G

C

C

G

G

C

C

G

A

T

P

P+ 5’−Phosphat

• Polymerisation (completion)..........A G C

G

G

C

C A A

T

G

C

C

G

G

C

• PCR (polymerase chain reaction)... ........ duplicate strands

Separation and analysis of DNA strands

• Affinity purification (sep. by biotin)..A C

G

C

G

T

A

A

T

G

C TC

B T

A

A

T

T

A + 5’−Biotin TC

B T

A

A

T

T

A

• Gel electrophoresis (sep. by length) sort and detect strands

• Sequencing (readout).......................A C G G A A C 5’−ACGGAAC−3’



NP−hard Knapsack Problem

Algorithm for Solution to the

3.



Knapsack ProblemNP-hard decision problem, exponential need of resources

DefinitionGiven n natural numbers a1, ...,an and reference number b ∈ NIs there a subset I ⊆ {1, ...,n} with

∑i∈I

ai = b ?

Explanationa1, ...,an: weights of objects 1, ...,n.Is there a possibility to pack a selection of these objects intothe knapsack which exactly meets the reference weight b?

Example

.

9 95

7 7

2

2

a = 7a = 2b = 9

1

2

3

a = 5

?object 2

object 1

object 3 "yes"solution



Idea of Wetware Solution to Knapsack Problem

Brute force approach• Encode a1, . . . ,an into DNA double strands by length (c ·ai )• Generate all solution candidates by a controlled

split-and-combine strategy• Separate final DNA pool by length• Detect DNA at Starter length + c · b and answer yes

by lengthseparation

10

20

30

5

15

25

35

a1

Starter

Starter

a1Starter Starter

Starter

Starter

Starter

a1

a1

a2

a2

a2

Starter

Starter

Starter

Starter

Starter

Starter

Starter

Starter

a1

a2

a1

a2

a1 a2

a1 a2

DNA operationsDNA operations DNA operations

a3

a3

a3

a3

a3

M. Sturm, T. Hinze. Verfahren zur Ausführung von mathematischen Operationen mittels eines DNA-Computers undDNA-Computer hierzu. Deutsches Patent DE 101 59 886 B4, IPC G06N 3/00, erteilt 2010



Split-and-combine Strategydoubles number of combinations by addition of one object

Union

Ligation

Split

Union

DNA pool composed of combinations from k objectsstart with k = 0 (starters only)

DNA pool composed of combinations from k+1 objects

Labelling 5’+Padd object a_(k+1)

Starter

Starter

Starter

Starter

a1

a1

a2

a2

a2

Starter

a1Starter

a2

a2

a1Starter

Starter



Simple Implementation of a Problem Instance• n = 3 objects taken from plasmid (pQE30 cleaved with PvuII)

• a1 = 719, a2 = 393, a3 = 270, b = 1112, c = 1

• Exponential need of resources moved from time to space

• Final sequencing of DNA band corresponding to b reveals “yes”

• Limited scalability of the algorithm due to side effects andamount of DNA

100

1000

1200

1

719 bp

393 bp

270 bp

750 bp

500 bp

250 bp

150 bp

bp

270270

393393

663393 270

719719

719 270

719 393

719 393 270

989

1112

1382

bp2 1 2 3



Side Effects and Perturbationsprevent DNA operations from running in an ideal manner

• Loss of DNA• Incomplete reactions• Non-specificities• Malformed DNA (artefacts)• DNA damage• Contaminations or impurities of DNA solutions• . . . (many others)

Coping with side effects is a hard challenge to overcome inpractical in-vitro DNA computation. Assuming an error rate of5% per operation and having a sequence of 10 operations, theoverall success rate is merely 0.9510 · 100 ≈ 60%.



Trends, and Perspectives

Further Applications,

4.



Milestones of DNA-based Computing• Pioneering era after Adleman’s experiment

• Refinement and improvement of techniques and encodingschemes complemented by much theoretical work

• Addressable DNA-based memory able to store data from files

• Computing by DNA self-assembly promising clue towardsfreely programmable nanomachines

Adleman’s

DNA−based solution to

Hamiltonian path problem

(n = 7 nodes)

www.usc.edu

1994


satisfiability problem

by brute force approach

(n = 20 variables)

www.sciencemag.org

2002







Adleman’s



(n = 7 nodes)

www.usc.edu

1994




(n = 20 variables)

www.sciencemag.org

2002

DNA chips for

massive parallel computing

by string matching

approx. 40,000 spots

2010

www.newscientist.com







Adleman’s



(n = 7 nodes)

www.usc.edu

1994




(n = 20 variables)

www.sciencemag.org

2002

DNA chips for

massive parallel computing

by string matching

approx. 40,000 spots

2010

www.newscientist.com

algorithmic self−assembly

for cellular automaton and

universal DNA computer

2014

PLoS Biology



International Molecular Computing Community

≈ 500 researchers worldwide, conference series like CMC, DNA, UC, . . .In-vitro Molecular Computing Based on DNA Strands Thomas Hinze


Take Home Message

Living organisms comprise almost perfectDNA-based computers. We are going to learn

and to adapt the underlying principles forutilisation in vitro. There are first successes but

there is still a lot of work to do.

Further and more detailed information

• T. Hinze, M. Sturm. Rechnen mit DNA -Eine Einführung in Theorie und Praxis.De Gruyter, eBook, 2014

• T. Hinze. Computer der Natur.bookboon.com, eBook (for free), 2013



Take Home Message

Living organisms comprise almost perfectDNA-based computers. We are going to learn

and to adapt the underlying principles forutilisation in vitro. There are first successes but

there is still a lot of work to do.

Further and more detailed information

• T. Hinze, M. Sturm. Rechnen mit DNA -Eine Einführung in Theorie und Praxis.De Gruyter, eBook, 2014

• T. Hinze. Computer der Natur.bookboon.com, eBook (for free), 2013


Documents

In-vitro Molecular Computing Based on DNA Strandshinze/hinze-slides... · 2015-09-15 · In-vitro Molecular Computing Based on DNA Strands An unconventional computing concept inspired