• Investment World
• eWorld
• Brand Line
• Mentor
• Life
• Canvas
• Praxis
• Urban Pulse
• Brand Quest
• The New Manager

Stocks

• Quotes
• SE Diary
• Scoreboard
• Open-End Mutual Fund

Cross Currency

• Rates
  
  

Shipping

• Ports
  
  

Archives

• Yesterday
• Datewise
• Resources
• In Focus
• In Depth

Group Sites

• The Hindu
• The Hindu ePaper
• Business Line
• Business Line ePaper
• Sportstar
• Frontline
• The Hindu eBooks
• The Hindu Images

Dr S. Chellaiah

The first part of this article, “IT is in the DNA too”, appeared in eWorld dated November 12. Here, in the second and concluding part, we look at the differences between a DNA computer and the traditional one, and the applications of a DNA computer. Since, presently, computers use chips made out of silicon, for this discussion, we shall call the traditional computer a ‘silicon computer’.

Architecture: Silicon computers use the von Neumann architecture. This involves a repetitive process of “fetch and execute” referring to the retrieval of the value of a variable value from memory, performing calculations and then retrieving the next variable and repeating this process. DNA computing follows the stochastic, massively parallel processing architecture. It is this ability to work in parallel that is appealing. This feature makes DNA computers best suited for attacking NP-complete problems. (Refer Note 1 in part one of this article.)

Composition: DNA computers use biological molecules to denote alphanumeric characters whereas silicon computers use electrical charge.

Representation: DNA computers use base-four system, in the sense, there are four nitrogenous bases A, T, C and G for representation. Silicon computers use base two to represent numbers and symbols. Thus, there can be a zero or one.

Speed: A single DNA molecule works at the speed of about one thousandth (0.001) million instructions per second whereas supercomputers operate at speeds higher than 1,000 million instructions per second.

Storage: DNA has immense storage capability. To store 1 bit of information, one cubic nanometer of DNA is required whereas 10 to the power of 12 cubic nanometers of space is needed for a traditional (silicon) computer. The density of information storage is trillion times more in DNA.

Power consumption: With 1 joule of energy, 2 x 10 to the power of 19 operations can be performed on a DNA computer as opposed to 10 to the power of 9 operations on a silicon computer. The energy (and cost) needed for extracting the information is not included in the above figures.

Others: In DNA computers, post processing is needed to interpret the DNA strand and infer the answer. In silicon computers, the answer is displayed on a monitor or printed on a paper or stored in a disk.

Amount of DNA: The DNA can store a lot of information. But the amount of DNA required for calculations also increases exponentially. Estimates reveal that to solve the Hamiltonian Problem with 200 cities, the amount of DNA required will exceed the weight of the earth.

Post processing: The interpretation of the results is a long, cumbersome process. The calculations proceed fast but one must analyse the resulting DNA strand to find the result.

Synthesis: The accuracy of the process is a limiting factor. Synthesising large quantities of DNA strands will include mismatching pairs. The quality of the enzymes also plays a critical role.

Stability: Complicated calculations require big molecules. With increase in size, the probability of the molecules being sheared is also high.

Shelf life: All DNA have half life periods. If the solution is not used before that time, then DNA decays.

Errors: DNA computers are prone to errors. The algorithms used can result in the number of DNA strands, a) remaining constant through all steps or b) decreasing with each step or c) remaining constant for a few steps and then decreasing. During extraction, only about 95 per cent of those DNA that match a pattern will be extracted. This error is called Type 1 or ‘false negative’. Strands that do not match the pattern may also get extracted. This error is called Type 2 or ‘false positive’ and occurs 1 in a million.

In general, for sequential calculations, DNA computers are no match for silicon computers.

The applications of DNA computing can be categorised into two types:

Those that require computation (solving mathematical equations, number crunching, etc) such as Travelling Salesman Problem in which the computer only facilitates the solution of a set of equations that cannot be solved manually. This is called mathematical computing.

Those that use DNA to compare/investigate and identify the presence of abnormalities and accordingly trigger an action. In this case, the DNA primarily functions as a sensor with an ability to act based on the sensed signal. This is different from mathematical computing.

In November 2001, researchers at Weizmann Institute, Israel, built a DNA computer that worked without any human supervision (like automaton). It used DNA as input and two enzymes as the hardware, both of them dissolved in a liquid. The computer was able to determine whether an input string had an even number of zeroes or not. The answer was found by running the liquid with the resulting DNA molecule through an electrophoretic gel. This computer was so tiny that a trillion of them could be fitted into a drop of water!

In August 2003, Milan Stojanovic of Columbia University in New York and Darko Stefanovic of University of New Mexico in Albuquerque demonstrated MAYA, a DNA computer that played tic-tac-toe. The board for tic-tac-toe consisted of three square (1-cm square) wells. There were nine unique DNA strands with each representing a well. The player (person) indicated his/her choice of square by dropping DNA that represented that square into all the wells (as all wells must become aware of the choice). The computer used DNA and enzymes to determine where it should place its cross (or nought) and indicated its choice by a green glow. This DNA computer could be programmed to always win or lose.

In Vivo Computing refers to the use of these DNA computing principles as a “DNA Doctor” inside the human body. Can human cells be injected with DNA to detect if a cell is cancerous or not and if cancerous, trigger subsequent actions like medication? An interesting thought indeed!

RNAi is a process called RNA interference in which small molecules of RNA stop the production of a protein by a DNA (As most readers know, RNA denotes a large biomolecule similar to DNA). In human cells, RNAi is already present. In the body, molecules called short interfering RNA (siRNA) recognise certain DNA in genes and then shut them down. In May 2007, Scientific American and Nature Biotechnology reported that Harvard scientists have developed siRNAs to identify a fluorescent protein in a gene and switch off the production of the protein in our body.

Scientists in Stanford and Princeton have proposed to use the immense parallel processing power of DNA to crack the information encrypted using the US Government’s digital encryption standard (DES). To decrypt by brute force (trying different million /billion possible combinations) will take many millennia. But if those combinations are tried in a parallel manner, then decryption becomes easy and quick. It is like having billion personal computers working simultaneously with each trying one possible combination.

DNA chip (also known by other names — DNA microarray, gene chip, gene array, and genome chip) is not the chip that is referred to in the microelectronics field (integrated chips used in computers).

These are merely collections of DNA spots denoting single genes in the form of an array on any suitable base material — such as glass, plastic or silicon. The purpose is to identify (qualitative) and/or calculate the large amount of genes simultaneously. Thus, in the strict sense, this does not refer to DNA computing.

In July 2002, Olympus Optical Co revealed a DNA computer to do gene analysis. It had two parts — Molecular calculation component and Electronic calculation component. What took about three days was completed in six hours using this computer. Today, there are a few companies that manufacture DNA chips.

Do we bid adios to silicon? Not really. Silicon computers will continue to rule the land for a long time. DNA computers are good at analysing multiple solutions simultaneously but to perform a single calculation, silicon computers are the best. Also, the post processing to find the solution from DNA is long and laborious. (Incidentally, RNA has also been used for computing — refer Note.) Hence, the commercialisation of DNA computers is far away.

Some predict that a hybrid computer may use both silicon and DNA to capitalise on the best of both worlds. Well may be!

As Hawksett from the Guinness World Records remarked, “DNA computing is an area of research that leaves the science fiction writers struggling to keep up”.

And while science fiction writers struggle, thanks mom, but for your casual remark, I would not have been exposed to an interesting and nascent technology.

The author is Consultant, Satyam Computer Services Ltd, and can be reached at chellaiah_s@satyam.com

Note: RNA has been found to be better than DNA and in January 2000, scientists in Princeton University used it to solve the ‘Knight Problem’.

Stories in this Section
Put some zing into SMS

Copyright © 2007, The Hindu Business Line. Republication or redissemination of the contents of this screen are expressly prohibited without the written consent of The Hindu Business Line