DNA Computing Computation with the Code of Life

1
DNA ComputingComputation with the Code of Life

• Michael Ang
• ltm_at_michaelang.comgt
• Interactive Telecommunications Program
• New York University
• February 16, 2007

2
What is it?

• Invented by Len Adleman (1994)
• Realized combinatorial properties of DNA could be
  used to solve problems encoded as strings of DNA

3
Why is it interesting?

• Cross-discipline (CS meets Molecular Biology)
• High data density
• Data and computation happens at molecular level
• Each base of DNA is 0.35nm
• Massively parallel
• 1012 or more copies of DNA in a test tube
• Use biological enzymes to make these copies
• Energy efficient
• Potential to perform computation inside the body
• Though big challenges before a reality

4
What is DNA?

• Source code to life
• Instructions for building and regulating cells
• Data store for genetic inheritance
• Cellular machinery (enzymes) translates DNA into
  proteins, duplicates, repairs, etc.
• Think of enzymes as hardware, DNA as software

5
What is DNA made of?

• Composed of four nucleotides ( sugar-phosphate
  backbone)
• A Adenine
• T Thymine
• C Cytosine
• G Guanine
• Bond in pairs
• A T
• C G

6
How does it work?

• Use specially coded DNA as initial conditions for
  biological reaction
• Natural enzymes duplicate DNA
• Matching DNA base pairs attach to each other
• Find answer in resulting soup of DNA strands

7
Travelling Salesman

• Simple problem (at small scale)
• Complexity scales exponentially

8
Algorithm

1. Generate all possible routes
2. Select routes that start with the initial city
   and end with the destination city
3. Select itineraries with the correct number of
   cities
4. Select itineraries that contain each city once

9
Algorithm

1. Generate all possible routes
2. Select routes that start with the initial city
   and end with the destination city
3. Select itineraries with the correct number of
   cities
4. Select itineraries that contain each city once

10
City Encoding
11
Annealing

• Use a lot (1013) copies of each city and path

12
Weve made all the routes
13
Algorithm

1. Generate all possible routes
2. Select routes that start with the initial city
   and end with the destination city
3. Select itineraries with the correct number of
   cities
4. Select itineraries that contain each city once

14
Polymerase Chain Reaction
15
And repeat
16
Algorithm

1. Generate all possible routes
2. Select routes that start with the initial city
   and end with the destination city
3. Select itineraries with the correct number of
   cities
4. Select itineraries that contain each city once

17
Gel electrophoresis

• Race the DNA through a gel
• Shorter lengths go faster

18
Algorithm

1. Generate all possible routes
2. Select routes that start with the initial city
   and end with the destination city
3. Select itineraries with the correct number of
   cities
4. Select itineraries that contain each city once

19
Affinity purify

• Pull out strands matching one city at a time

20
We have our answer

• Sequence result or perform more PCR reactions to
  determine combinatorically
• Took Adleman 7 days to run through the steps
• Actual computation took minutes

21
Problems/Challenges

• Relatively high error rate in DNA
• Sometimes bases dont align properly
• Problem complexity still scales exponentially
• 200 city problem might take DNA weighing more
  than Earth
• Many people doubtful

22
Can I do it?

• If you have access to a lab, yes
• DIY seems possible (lt500)

Personal Biocomputing Eugene Thacker
23
Further Research

• Use to break DES (1995)
• Turing Machine (at least similar to) (2001)
• Play tic-tac-toe (2003)
• DNA computer to detect and treat cancer (2004)
• Use enzymes and DNA to create state machine using
  mRNA as input
• Output DNA sequence to suppress disease causing
  gene
• In vitro (petri dish) so far
• Much harder in vivo (living cell)

24
Maya II Tic Tac Toe
25
Maya II Logic Gates
26
Maya II Fluorescent Output
27
DNA Automata
28
A nanoscale programmable computing machine with
input, output, software and hardware made of
biomolecules Nature 414, 430-434 (2001)

• 1012 automata run independently and in parallel
• on potentially distinct inputs
• in 120 ml
• at room temperature
• at combined rate of 109 transitions per second
• with accuracy greater than 99.8 per transition,
• consuming less than 10-10 Watt.

29
(Video from Weizmann Institute)
30
So what?

• Computation at molecular level
• Computation (potentially) within cells
• Operate with cellular messages as input and
  output
• Operate near theoretical power limits
• Computer Science implemented in Biology
• Practical applications still a dream

31
To correctly gauge the practicality of molecular
computing will require inputs from experts in a
wide variety of fields including biology,
chemistry, computer science, engineering,
mathematics and physics. - Len Adleman (1995)
32
References

• What is DNA - Genetics Home Reference
• DNA Computer Could Target Cancer Nanotech Web
• DNA Computing A Primer Arstechnica
• On Constructing a Molecular Computer Len
  Adleman
• Breaking DES using a molecular computer - D.
  Boneh, C. Dunworth, and R. Lipton
• Injectable Medibots Programmable DNA could
  diagnose and treat cancer - Alexandra Goho
• First game-playing DNA computer revealed New
  Scientist
• Programmable and autonomous computing machine
  made of biomolecules - Benenson, Paz-Elizur,
  Adar, Keinan, Livneh Shapiro
• Personal Biocomputing Eugene Thacker
• Biological Nanocomputer Weizman Institute
• Computer Made from DNA and Enzymes National
  Geographic News

33
Thanks

• Michael Ang
• ltm_at_michaelang.comgt
• Interactive Telecommunications Program
• New York University
• February 16, 2007