DNA-Based Computing & Development of Efficient Operators

for Solving Traveling Salesman Problem

Tai Hyun Park School of Chemical Engineering Seoul National University

Contents
Introduction DNA computing for solving traveling salesman problem
Traveling salesman problem (TSP) Molecular algorithm for TSP Experimental implementation

Unit operators in DNA computing

Denaturation temperature gradient polymerase chain reaction (DTG-PCR) Initial pool generation using parallel overlap assembly (POA)

Rise and Growth of DNA Computing

Adlemans work in 1994
Hamiltonian path problem (graph problem, NP problem) City and road information representation using DNA sequences (indicative information) Solution-based DNA computing

Lius work in 2000

SAT problem Surface-based DNA computing

Benensons work in 2004

Application to disease diagnosis and drug (antisense) administration

Our Work in DNA Computing Era

Solving traveling salesman problem
Graph problem with weighted edges Representation of numerical information Expansion to larger problems (in progress)

Development of unit operators

Denaturation temperature gradient PCR (DTG-PCR) Initial pool generation for TSP using parallel overlap assembly (POA) Modeling and simulation of DTG-PCR and POA

New application to medical diagnosis

Disease diagnosis model development Experimental verification (in progress)

Objective (traveling salesman problem)

Application of DNA computing to mathematical problems
7-city traveling salesman problem Graph problem with weighted edges

Representation of numerical values (weighted edges) Expansion to practical problems

26-variable TSP 26 subway stations in Seoul

Traveling Salesman Problem

Find
The cheapest way of visiting all the cities and returning to the starting point when a number of cities to visit and the traveling cost between each pair of cities are given.

Previous work for weight (cost) representation

DNA length DNA concentration

Our method for weight (cost) representation

Thermal stability of DNA duplex Melting temperature (Tm), GC content

Target Problem & Encoding Method

3

4 1 7-city traveling salesman problem 7-city traveling salesman problem

- -77cities (0 to 6), 23 roads, 55costs cities (0 to 6), 23 roads, costs
11

start & end city

3
3 7 5 3 3 3 9

11 3

- -optimal path: 01234560 optimal path: 01234560

6
3

Oligonucleotides & Encoding Oligonucleotides & Encoding

- -roads are 40-base ssDNA roads are 40-base ssDNA - -35 oligonucleotides 35 oligonucleotides (7 cities, 23 roads, 55costs) (7 cities, 23 roads, costs)

(20 bases)

(20bases)

(20 bases)

- -cities and costs are 20-base ssDNA 5 cities and costs are 20-base ssDNA

city A 3

cost A to B

city B 5

road A to B
(40 bases)

Weight (Cost) Encoding

20-base ssDNA 20-base ssDNA with 50% GC with 50% GC content content 20-base ssDNA 20-base ssDNA with higher GC with higher GC content content 20-base ssDNA 20-base ssDNA with lower GC with lower GC content content

Seoul

$ 500

Tokyo

$ 200

Sapporo

Road from Seoul to Tokyo

Road from Tokyo to Sapporo

40-base ssDNA 40-base ssDNA complementary complementary to city and cost sequence to city and cost sequence

Lower cost Lower cost

More amplification of More amplification of DNA strand which has DNA strand which has lower sum of cost lower sum of cost Separation Separation and detection and detection

DTG-PCR

NACST/seq

Lower Tm Lower Tm

TGGE

Denaturation Temperature Gradient Polymerase Chain Reaction (DTG-PCR)

Conventional PCR
Denaturation (Td) Annealing (Ta) Extention (Te)
Denaturation
92~95

Extension
68~72

Modification of conventional PCR protocol

Variation in Td

(Tm-5)

Annealing Cycle

Denaturation Temperature Gradient

Final denaturation temperature~ 94 High Td Amplification regardless of Tm

Initial denaturation Temperature ~ 70

Low Td More amplification of DNA strand which has lower Tm (lower cost) Cycle progress

Biased operator: more amplification of DNA strands with lower Tm biased search for lower cost

Molecular Algorithm
1 2

(1) Generation of all random paths Hybridization/ligation

(2) Selection of paths satisfying TSP conditions PCR/electrophoresis/affinity 3

(3) More amplification of the cheapest path DTG-PCR

Readout!
Sequencing

4
(4) Separation of the cheapest path TGGE

(5) Elution of the cheapest path Gel elution

Sequence Design for Cities and Costs

Using NACST/seq Non-cross hybridization Similar Tm among cities Different Tm among costs

Experimental Implementation for TSP Conditions

Strating and ending with city 0
PCR using primers complementary to city 0

Visiting every city

A series of affinity chromatography Each affinity column contains ssDNA complementary to each city.

Cheapest path
DTG-PCR

Experimental Results
Random path generation
(by hybridization and ligation)

Selective amplification of paths starting and ending with city 0

(by PCR using primers complementary to city 0)

M 1

300bp M: 50 bp ladder lane 1,2: after hybridization/ligation lane 3: mixture of ssDNA

Separation of paths containing every city

(by a series of affinity chromatography)

More amplification of paths with lower costs

(by DTG-PCR)

M
streptavidin biotin

Magnetic Bead
300bp (8 cities)

City 0

Cost

City 1

Cost

City 2

Cost

Separation of the path with lowest cost

(by TGGE)

TGGE (Temperature Gradient Gel Electrophoresis)

3
DN A mo le c u le

major band

Readout
(by cloning and sequencing)
TTCTGCGTTGTTTCGGGGTACAGTGGCTCCTCCGTT CCGCCTGCACTGTGGAGAGGTGAGCAGTGGCTCCTCCGTT CCGCGTGGATTCACAAGGCCATCGCAGTGGCTCCTCCGTT CCGCATACGGCGTGGTTTTTCGGGCAGTGGCTCCTCCGTT CCGCAAACGGTCGTAAGTGATGAACAGTGGCTCCTCCGTT CCGCGCACAGTCCACCTGTAGACACAGTGGCTCCTCCGTT CCGCTATGTCCAGCTGTCGCAAAGCAGTGGCTCCTCCGTT CCGCTTCTGCGTTGTTTCGGGGTA

City 0 Cost 3 City 1 City 2 City 3 Cost 3 Cost 3 Cost 3

City 4 Cost 3 City 5 City 6 City 0 Cost 3 Cost 3

Toward Larger Problems

4 1 2 1 1 1 5 1 2 1 1 1 1 2 4 3 2 2 1 3 2 5 4 2 2 2 2 2 3 3 2 1 4 2 2 8 1 1 7 3

2 2 1

4 7 11

Subway routes in Seoul Subway routes in Seoul

Target Problem: 26-City TSP

1 1 1 1 24 9 18 19 5 11 1 1 1 2 1 1 2 25 4 2 8 232 22 3 4 2 3 2 2 15 1 13 5 4 16 2 7 2 2 2 2 2 3 7 12 10 17 21 14 3 2 1 1 4 2 2 11 8 4 20 1 7 1 1 3 3 6 5 4 2

Graph with 26 vertexes (cities) and 92 edges (roads)

Vertex: station connected with more than two stations Weight: number of stations between vertex stations

Initial Pool Generation with POA

Initial pool
A combinatorial library that contains numerical or indicative information

Initial pool generation

Prerequisite step of most molecular algorithms for mathematical problems

Initial pool generation methods

Hybridization and ligation method Parallel overlap assembly method

Initial Pool Generation for TSP

: POA vs. Hybiridzation/Ligation
Oligonucleotides

First cycle extension Phosphorylation Hybridization: slow cooling Second cycle extension

Ligation: ligase

Repeat of denaturation/annealing /extension steps

Experimental Results
M
900 700 500 300 200 100

Lane Lane Lane Lane

M: 50 bp ladder 1: mixture of ssDNA 2: hybridization/ligation 3: POA

Quantification by image analysis Quantification by image analysis

Marker

Lane 1

Lane 2

Lane 3

900 700 500 300 200 100

Comparison of Two Methods

Parallel overlap assembly
Population size is almost preserved throughout the process. No need of 5-phosphorylation Economy Less time & reagents consuming More time & reagents consuming

Viewpoint

Hybridization/ligation
Population size decreases as hybridization proceeds.

Scalability

5-phosphorylation is necessary

High error rate by non-specific priming

Fidelity

Low error rate (higher hybridization specificity)

Summary
Development of a molecular algorithm for TSP based on the melting temperature difference Development of DTG-PCR as an efficient operator for the graph problem with weighted edge Success in solving 7-city TSP 26-city traveling salesman problem POA as an efficient operator for large size TSP

Acknowledgments
Prof. Byoung-Tak Zhang (School of Computer Science and Engineering, SNU) Ji Youn Lee (School of Chemical Engineering, SNU) Soo-Yong Shin (SNU) (School of Computer Science and Engineering, SNU) The Ministry of Commerce, Industry, and Energy