REVIEW

Chinese Science Bulletin 2006 Vol. 51 No. 12 1413—1420 invented and this might lead to a new revolution in in-
DOI: 10.1007/s11434-006-2012-5 formation science. On such background, DNA cryp-
tography is developed with the research of DNA com-
New field of cryptography: puting (also called biological computing or molecular
computing). The traditional cryptography made great
DNA cryptography progress in the 20th century with the development of
electronic technology and is widely used currently.
XIAO Guozhen1, LU Mingxin1, QIN Lei2 Quantum cryptography was invented in 1970s and has
& LAI Xuejia3 made some progress in recent decades, but there is still
some distance from being used in practice. DNA cryp-
1. National Key Lab of ISN, Xidian University, Xi’an 710071, China;
2. Queen’s University, Cancer Research Institute, 10 Stuart St. Kingston,
tography drawn attention after DNA computing was
ON K7L3N6, Canada; first proposed by Adleman in 1994, and now it has be-
3. Department of Computer Science & Engineer, Shanghai JiaoTong come the frontier of cryptography. DNA cryptography,
University, Shanghai 200030, China traditional cryptography and quantum cryptography are
Correspondence should be addressed to Lu Mingxin (email: seulmx@
126.com ) developed under the same aim-information security, but
Received December 19, 2005; accepted January 16, 2006 with quite different ways, and all the three kinds of
cryptography might constitute the major fields of future
Abstract DNA cryptography is a new born cryp-
cryptography. In this paper, the biological background,
tographic field emerged with the research of DNA
research progress and prospects of DNA cryptography
computing, in which DNA is used as information car-
are reviewed and discussed to highlight its future stud-
rier and the modern biological technology is used as
ies.
implementation tool. The vast parallelism and
extraordinary information density inherent in DNA
1 Biological background
molecules are explored for cryptographic purposes
such as encryption, authentication, signature, and so 1.1 DNA
on. In this paper, we briefly introduce the biological
DNA is the abbreviation for deoxyribonucleic acid
background of DNA cryptography and the principle of
which is the germ plasm of all life styles. DNA is a
DNA computing, summarize the progress of DNA
cryptographic research and several key problems, kind of biological macromolecule and is made of nu-
discuss the trend of DNA cryptography, and compare cleotides. Each nucleotide contains a single base and
the status, security and application fields of DNA there are four kinds of bases, which are adenine (A) and
cryptography with those of traditional cryptography thymine (T) or cytosine (C) and guanine (G), corre-
and quantum cryptography. It is pointed out that all sponding to four kinds of nucleotides. A single-stranded
the three kinds of cryptography have their own ad- DNA is constructed with orientation: one end is called
vantages and disadvantages and complement each 5′, and the other end is called 3′. Usually DNA exists as
other in future practical application. The current main double-stranded molecules in nature. The two comple-
difficulties of DNA cryptography are the absence of mentary DNA strands are held together to form a dou-
effective secure theory and simple realizable method. ble-helix structure by hydrogen bonds between the
The main goal of the research of DNA cryptography is complementary bases of A and T (or C and G). The
exploring characteristics of DNA molecule and reac- double-helix structure was discovered by Watson and
tion, establishing corresponding theories, discovering
Crick; thus the complementary structure is called Wat-
possible development directions, searching for sim-
son-Crick complementarity[1]. Their discovery is one of
ple methods of realizing DNA cryptography, and lay-
ing the basis for future development. the greatest scientific discoveries of the 20th century
and reduced genetics to chemistry and laid the founda-
Keywords: cryptography, DNA cryptography, DNA computing.
tions for the next half century of biology[2].
The vast parallelism, exceptional energy efficiency 1.2 PCR and DNA chip
and extraordinary information density inherent in DNA
molecules are being explored for computing, data stor- Within decades, great progress has been made in
age and cryptography. In such research area, novel gene biotechnology. With invention and maturation of
computers, data storage and cryptography might be these technologies, such as DNA synthesis, PCR, elec-

www.scichina.com www.springerlink.com 1413

REVIEW
trophoresis, DNA sequencing hybridization and other In 1994, Adleman demonstrated the first DNA
techniques, scientists can synthesize, amplify, isolate, computing, which marked the beginning of a new stage
digest and sequence DNA strands very easily. It opens in the era of information[8]. In the following researches,
the door to new application of DNA different forms in scientists find that the vast parallelism, exceptional en-
the evolution of life. PCR technology is used in all ex- ergy efficiency and extraordinary information density
isting DNA computing methods and DNA cryptography are inherent in DNA molecules[8 12]. In 2002, a team
－

methods, while the efficiency of DNA hybridization is led by Adleman solved a 3-SAT problem with more
improved greatly with DNA chip technology. These than 1 million possibilities on a simple DNA computer
two important technologies will be briefly introduced in after an exhaustive searching[13]. In 2005，it is declared
this paper.
that a team led by Ehud Keinan invented a biomolecu-
Polymerase Chain Reaction （PCR） was invented in
lar computer that used little more than DNA and en-
1983. It is one of the most important invention in mod- zymes could perform a billion operations simultane-
ern biology[3]. Since the DNA molecule is tiny in vol-
ously[14]. Adleman reviewed DNA computing as fol-
ume, it is difficult to manipulate a small quantity of
lows: “For thousands of years, humans have tried to
given DNA directly, while it will be quite easy to ma-
enhance their inherent computational abilities using
nipulate a great mount of DNA after amplification.
manufactured devices. Mechanical devices such as the
PCR is a fast DNA amplification technology based on
abacus, the adding machine, and the tabulating machine
Watson-Crick complementarity. Two complementary
were important advances. But it was only with the ad-
oligonucleotide primers are annealed to double-
vent of electronic devices and, in particular, the elec-
stranded target DNA strands, and the necessary target
tronic computer some 60 years ago that a qualitative
DNA can be amplified after a serial of polymerase re-
action from 5′ to 3′ with the aid of polymerase enzyme. threshold seems to have been passed and problems of
The PCR is a very sensitive method, and in theory a considerable difficulty could be solved. It appears that a
single target DNA molecule can be amplified to 106 molecular device has now been used to pass this quali-
after 20 cycles. Thus one can effectively amplify a lot tative threshold for a second time.”[13] Scientists have
of DNA strands within a very short time [3]. also made progress in the theory of DNA computing
DNA chip is also known as DNA microarray or gene and explored several feasible computing models, such
chip or oligonucleotide chip or biological chip, which is as the model used by Adleman in 1994[8]. Here it is
fabricated with in situ synthesized oligo nucleic acids called Hamiltonian path model, and the model based on
or spotted cDNA probes according to published meth- DNA chip[14,15] and the sticker model proposed by
ods of Fodor and Brown[4 7]. Tens of thousands, even
－ Adleman[16]. Below, the widely used Hamiltonian path
millions of DNA probes are arranged in a square area model and sticker model are briefly introduced.
less than 1 square inch on glass or silicon matrix. And 2.1 Hamiltonian Path Model
as their counterparts, numerous labelled probes are
used to anneal with probes on the chip to get various In 1994, Adleman used DNA computing to solve an
hybridization spectrums revealing genetic information. instance of the directed Hamiltonian path problem[8].
Thus the hybridization efficiency can be raised thou- The computing model proposed by Adleman is also
sands of times and even more. used by Lipton to solve SAT problems[17]. The Hamil-
tonian path problem is to find a path that begins at vin,
2 DNA computing ends at vout and enters every other vertex exactly once
The development of DNA cryptography benefits on a directed graph. For each vertex i in the graph, a
from the progress of DNA computing (also called mo- random 20-mer oligonulecotide (short DNA strand) Oi
lecular computing or biological computing). On the one was generated. Here, mer is the long measure of oli-
hand, cryptography always has some relationship with gonulecotide. The following O2, O3 and O4 denote
the corresponding computing model more or less. On vertices 2, 3, and 4, respectively. All the following oli-
the other hand, some biological technologies used in gonucleotides are written from 5′ to 3′.
DNA computation are also used in DNA cryptography.
O2 = TATCGGATCGGTATATCCGA,
For these reasons, DNA computing is briefly intro-
duced here. O3 = GCTATTCGAGCTTAAAGCTA,

1414 Chinese Science Bulletin Vol. 51 No. 12 June 2006

 REVIEW
O4 = GGCTAGGTACCAGCATGCTT .
For each edge i → j in the graph, an oligonuleco-
tide strand is derived from the 3′ 10-mer of Oi and from
the 5′ 10-mer of Oj. For each vertex i in the graph, Oi
is the Watson-Crick complement of Oi. All the follow-
Fig. 2. Ligation reaction.
ing oligouncleotides in Fig. 1 are written as 5′ to 3′ ex-
cept O3 . was performed to amplify paths which started from vin
and ended at vout. Then oligonucleotides with correct
length were separated out by gel- electrophoresis. Paths
which did not entry some vertex were separated by a
biotin-avidin magnetic beads system. If there are DNA
strands left, it will show that the Hamiltonian path ex-
ists.
2.2 Sticker model
Both the sticker model and Hamiltonian path model
are based on Watson-Crick complementary. The dif-
ference is that in Hamiltonian path model, there are
short strands at beginning and long strands are formed
by way of anneal as the answer, while in sticker model,
Fig. 1. Directed graph. there are long single-stranded DNA at the beginning
and short stickers are annealed to long strands. The
O2→3 = GTATATCCGAGCTATTCGAG, sticker model is used both in refs. [13, 18]. The sticker
O3→ 4 = CTTAAAGCTAGGCTAGGTAC , model mainly includes memory structure and four op-
erations of combination, separation, setting particular
O3 = CGATAAGCTCGAATTTCGAT . bits and clearing bits. The memory structure of sticker
In the experiment, for each vertex i (except the vin model is shown in Fig. 3. More details of the sticker
and the vout) in the graph and for each edge i → j in model are shown in ref. [16].
the graph, sufficient Oi and Oi → j were mixed to- Despite the successes of the DNA computing in the-
ory and practice, it is still far away from discovering all
gether in a single ligation reaction. After a “graduated secrets in a cell, and thus, it is very likely that new
PCR” reaction, a lot of oligouncleotieds which denote computing models will be found in the future study.
different paths were generated. Note that the Oi oli- Just as Gifford said,“Transcriptional control and other
gonucleotides served as splints to bring oligonucleo- gene regulation mechanisms certainly play a paramount
tides associated with compatible edges together for role in the programming of cell behavior, but there may
ligation (Fig. 2). be other computational mechanisms lurking behind
After all steps of the reaction were finished, a PCR seemingly simple biological processes”[19].

Fig. 3. Sticker model.

www.scichina.com www.springerlink.com 1415

REVIEW
3 Progress and main problems in DNA cryptog- and has great potential utility value. Their method
raphy might be effective for solving the storage problem of
one-time pad. The disadvantage is that it is difficult to
3.1 Progress
prepare a huge DNA one-time pad in which data can be
DNA cryptography is a new born cryptography, in easily separated and read out. For the sender and re-
which DNA is used as information carrier and the ceiver, they have to do complex biological experiments,
modern biological technology is used as implementa- which can only be done in a well equipped lab and too
tion tool, and the vast parallelism, exceptional energy expensive to achieve. In many years, the scheme is not
efficiency and extraordinary information density in- feasible for the above reasons.
herent in DNA molecules are explored for crypto- For DNA steganography, Clelland et al. successfully
graphic purposes such as encryption, authentication, hid the famous “June 6 invasion: Normandy” in DNA
signature, and so on. Though DNA cryptography bene- microdots[21]. Their method is as follows:
fits from the research of DNA computing; it is not a 1) Encoding rule. A novel encoding method is pro-
direct result after all. The DNA computing proposed by posed instead of the traditional binary encoding. Nu-
Adleman cannot be regarded as DNA cryptography cleotides are used as quaternary code and each letter is
directly. In DNA computing, DNA technology is used denoted by three nucleotides. For example, the letter A
to solve difficult computational problems, while in is denoted by CGA, the letter B is denoted by CCA,
DNA cryptography, different difficult biological prob- etc.
lems are studied and used as the secure foundation of 2) Synthesizing secret-message DNA. The secret
DNA cryptosystems. The encryption process and the message is encoded into DNA sequence according to
decryption process can be regarded as computation. But the above code. For instance, AB is encoded as
not all DNA computations are involved in cryptography. CCGCCA. After coding, they synthesized a secret-
Note that DNA cryptography is not the same as genetic message DNA oligodeoxynucleotide containing an en-
code. Genetic code deals with the heredity of life and coded message 69 nucleotides long flanked by forward
belongs to the field of gene biotechnology. The re- and reverse PCR primers, each 20 nucleotides long.
search of DNA cryptography is at the beginning cur- Thus, the secrete-message DNA is prepared.
rently. There are few effective schemes[20 22]. However
－
3) Hiding message. They prepared concealing DNA
in this paper, two early representative DNA crypto- that is physically similar to the secret-message DNA by
graphic schemes are introduced. The first one shows sonicating human DNA to roughly 50 to 150 nucleotide
the extraordinary information density of DNA, but this pairs (average size) and denaturing it. The se-
scheme is very difficult to achieve. The second one is cret-message DNA and concealing DNA were mixed
much easier to be achieved, in which PCR is used for and attached on a piece of paper using common adhe-
decryption. The method of read data by PCR not only sives to form colorless microdots. Then the paper con-
has relationship with DNA computing but also is taining microdots can be posted by general mail ser-
widely applied in following studies. Though the second vice.
method is also called DNA steganography method, it is 4) Read. The sharing secrets for the sender and the
more appropriate to look on it as a DNA cryptography receiver are encoding rule and primers. After the re-
method according to its security foundation. ceiver gets the paper, he can easily find the microdots.
Gehani et al. achieved a one-time pad based on Since the intended receiver had gotten the primers and
DNA[20]. They argued that current practical applications encoding rule through a secure way, he could amplify
of cryptographic systems based on one-time pads are the secret-message DNA by perform PCR on DNA
limited to the confines of conventional electronic media microdots, sequence it and retrieve the message (plain-
whereas DNA is of extraordinary information density. text) according to the encoding rule.
A gram of DNA contains about 1021 DNA bases, or The basic process of the steganography method pro-
about 108 tera-bytes. Hence, a few grams of DNA may posed in ref. [21] is shown in Fig. 4. The secret-mes-
have the potential of storing all the data stored in the sage synthesizing process is shown in Fig. 4(a), the
world. Thus, DNA is very suitable to store a huge encoding rule is shown in Fig. 4(b), the PCR result is
one-time pad. The scheme proposed by Gehani et al. is shown in Fig. 4(c), the secret-message DNA and cor-
based on extraordinary information density of DNA responding message (plaintext) is shown in Fig. 4(d).

1416 Chinese Science Bulletin Vol. 51 No. 12 June 2006

 REVIEW

Fig. 4. Steganography method.

In fact, it is not correct that only the encoding rule is logical experiments have to be done in encryption step
considered key in Fig. 4. The true key should be prim- and decryption step, such as synthesizing message
ers and encoding rule. DNA strands, conducting PCR amplification and se-
quencing. Such experiments can only be done in a well
3.2 Main problems
equipped lab using current technology. For these rea-
The progress in the world shows that the research of sons, DNA cryptosystems are not convenient in prac-
present DNA cryptography is mainly confronted with tice and cannot compete with traditional cryptosystems.
the following problems: Luckily, modern biology has made much headway in
(i) Lack of the related theoretical basis. In 1949, recent twenty years. Many expensive experiments in
Shannon proposed the basic model and development the past have become routine experiments. With the
direction for modern privacy communication in his fa- further development of biology and new design of
mous paper “Communication theory of secrecy sys- DNA cryptosystem, the problems of “difficult and
tems”[23]. In the 1970s, it is proposed that the complex- expensive to realize” can be solved.
ity theory should be used as a powerful tool for design-
ing encryption algorithms, which also makes the emer- 4 Comparisons among DNA cryptography, tradi-
gence of public-key cryptosystem possible[24]. In the tional cryptography and quantum cryptography
following decades, new cryptosystems such as RSA,
EIGamal, DES and AES are invented[25 28]. Thus, the
－ 4.1 Development
traditional cryptography is perfected more and more. In Traditional cryptography can be traced back to Cae-
contrast, for DNA cryptography there is no mature cor- sar cipher 2000 years ago or even earlier. Related the-
responding theory. It is still an open problem as to what ory is almost sound. All the practical ciphers can be
are the model and security basis of DNA cryptography, seen as traditional ones. Quantum cryptography came
say nothing of the implementations. For lack of related into being in the 1970s, and the theory basis has been
theory it is difficult to design good DNA cryptographic prepared while implementation is difficult. By and
schemes. large, they have not been plunged into practical use.
(ii) Difficult to realize and expensive to apply. For DNA cryptography has only nearly ten years history,
the existing DNA cryptography schemes, many bio- the theory basis is under research and the application

www.scichina.com www.springerlink.com 1417

REVIEW
costs very much. Due to the vast parallelism, exceptional energy effi-
ciency and extraordinary information density inherent
4.2 Security
in DNA molecules, DNA cryptography can have spe-
Only computational security can be achieved for tra- cial advantages in some cryptographic purposes, such
ditional cryptographic schemes except for the one-time as secure data storage, authentication, digital signature,
pad, that is to say, an adversary with infinite power of steganography, and so on. DNA can even be used to
computation can break them theoretically. It is shown produce unforgeable contract, cash ticket and identifi-
that quantum computers have great and striking com- cation card.
putational potential[29 31]. Although there is uncertainty
－
Researches of all the three kinds of cryptography are
about the computational power of quantum computers, still in progress, and a great many problems remains to
it is possible that all the traditional schemes except for be solved especially for DNA and quantum cryptogra-
the one-time pad can be broken by using the future phy, this making it hard to predict the future. But from
quantum computers. Quantum cryptographic schemes the above discussions we think it is likely that they ex-
are unbreakable under current theories. Differently, ist and develop conjunctively and complement each
their security is based on Heinsberg's Uncertainty Prin- other rather than one of them falls into disuse thor-
ciple. Even if an eavesdropper is given the ability to do oughly.
whatever he wants, and has infinite computing re-
sources, so much as P=NP, it is still impossible to break 5 Development directions of DNA cryptography
such a scheme. Any behavior of eavesdropping will Since DNA cryptography is still in its immature
change the cipher so it can be detected. It is impossible stage, it is too early to predict the future development
for an adversary to obtain a totally same the quanta precisely. However, in view of the development of
with the intercepted one, thus the attempt to tamper but biological techniques and the requirement of cryptog-
without being detected in vain[32 36]. Therefore, quan-
－
raphy, we hold the following opinions:
tum key agreement schemes have unconditional secu- 1) DNA cryptography should be implemented by
rity. For the DNA cryptography, the main security basis using modern biological techniques as tools and bio-
is the restriction of biological techniques, which has logical hard problems as main security basis to fully
nothing to do with the computing power and immu- exert the special advantages.
nizes DNA cryptographic schemes against attacks us- Encryption and decryption are procedures of data
ing quantum computers. Nonetheless, the problem as to transform which, if described by mathematical methods,
what is the extent this kind of security and how long it are easier to be implemented than physical and chemi-
can be maintained it is still under exploration. cal ones in the present era of electronic computers and
the Internet. If other kinds of cryptosystems are neces-
4.3 Application
sary to be researched and developed, they should have
Traditional cryptosystems are the most convenient of properties such as higher security levels and storage
which the computation can be executed by electronic, density etc, which cannot be realized by electronic
quantum as well as DNA computers, the data can be computers by using mathematical methods. Thus, if
transmitted by wire, fiber, wireless channel and even by DNA cryptography is necessary to be developed, the
a messenger, and the storage can be CDs, magnetic advantages inherent in DNA should be fully explored,
medium, DNA and other storage medium. Using the such as developing nanoscopic storage based on the
traditional cryptography we can realize purposes as tiny volume of DNA, realizing fast encryption and de-
public and private key encryption, identity authentica- cryption based on the vast parallelism, and utilizing
tion and digital signature. Quantum cryptosystem is difficult biological problems that one can utilize but
implemented on quantum channels of which main ad- still far from fully understand them as the secure foun-
vantage lies in real-time communication. The disad- dation of DNA cryptography to realize novel crypto-
vantage lies in the secure data storage, which makes it system which can resist the attack from quantum com-
infeasible to implement public-key encryption and puters. Since it has not been made sure whether quan-
digital signature as easily as traditional one does. Under tum computers threaten the hardness of various
the current level of techniques, only by physical ways mathematical hard problems, these problems being se-
can the ciphertext of DNA cryptography be transmitted. curity basis cannot be excluded absolutely. Encryption

1418 Chinese Science Bulletin Vol. 51 No. 12 June 2006

 REVIEW
and decryption algorithms hard to be implemented us- It is more practical to make use of colony property of
ing electronic computers may be feasible using DNA plentiful DNA for cryptographer. For example, store
ones with regard to their vast parallel computational data by DNA chips and read data by hybridization,
ability. If these schemes withstand attacks by quantum which makes the operations of input/output faster and
computers, their computational security will be inher- more convenient. The method is easier to be imple-
ited into DNA schemes. Thereby, DNA cryptography mented than encoding message into nucleotides directly
does not absolutely repulse traditional cryptography while the storage density is somewhat lower.
and it is possible to construct a hybrid cryptosystem of 4) Currently, the main task for DNA cryptographers
them. is to establish the theory foundations and to accumulate
2) Security requirements the practical experience.
Regardless of the many differences between DNA It can be proved that there are vast parallelism, ex-
and traditional cryptography, they both satisfy the same ceptional energy efficiency and extraordinary informa-
characteristic of cryptography. The communication tion density inherent in DNA. This motivates the re-
model for DNA encryption is also made up of two par- search of DNA computing and cryptography. The cur-
ties, i.e. a sender and a receiver, which obtain the secret rent goal or difficulty is to find and make use of the
key in a secure or authenticated way and then commu- utmost potential, but the related research is in its initial
nicate securely with each other in an insecure or unau- stage. Sound theories have not been founded for both
thenticated channel. The security requirements should DNA computing and cryptography. Modern biology
also be founded upon the assumption proposed by
lays particular stress on experiments rather than theo-
Kerchoff that security should depend only on the se-
ries. There is no efficient way to measure the hardness
crecy of decryption key; that is, an attacker should be
of a biological problem and the security level of the
fully aware of all the details of encryption and decryp-
corresponding cryptosystems based on the problem. It
tion except the decryption key. It is under this assump-
is certainly urgent to find such a method similar to
tion that a cryptosystem can be said secure when any
computational complexity. Presently, the most impor-
attacker cannot break it[24]. More precisely, it must be
tant is to find the sound properties of DNA that can be
assumed that an attacker knows the basic biological
used to computation and encryption, to establish the
method the designer used, and has enough knowledge
and excellent laboratory devices to repeat the de- theoretical basis and to accumulate the experience,
signer’s operations. The only thing not known by the based on which the design of secure and practical DNA
attacker is the key. In a DNA cryptosystem, a key is cryptosystems is possible.
usually some substances of biological materials or a
6 Conclusions
preparation flow, and sometimes the experiment condi-
tions. The research of DNA cryptography is still at the be-
3) For DNA cryptography, the current research tar- ginning, and there are many problems to be solved. But
get should lie first in security and feasibility, second in the vast parallelism, exceptional energy efficiency and
storage density. extraordinary information density inherent in DNA
A sound cryptosystem should be secure as well as molecules endow DNA cryptography special advan-
easy to be implemented. The development of modern tages over other kinds of cryptography. Just as Adleman
biological technology makes it possible to express data said, “DNA computers, such as the one presented here,
by DNA, although the related research is just in its ini- illustrate that biological molecules (nucleicacids, pro-
tial stage. In fact, it is still difficult to operate the nano- teins, etc.) can be used for distinctly non-biological
scopic DNA directly. Scientists can easily operate purposes. For such purposes, these molecules represent
DNA with the aid of kinds of restriction enzymes only an untapped legacy of 3 billion years of evolution and
after DNA strands are amplified with amplification there is great potential in their further exploration”[13].
technology such as PCR. With the current technology,
it is also impossible to store all the worldwide data by Acknowledgements This work was supported by the Na-
using several grams of DNA. If the only requirement is tional Natural Science Foundation of China (Grant No.
to improve the density of storage, it is hard to imple- 60473028). We thank the review experts for their very valued
ment DNA cryptography at the present technique level. revise advice.

www.scichina.com www.springerlink.com 1419

REVIEW
References 19 Gifford D K. On the path to computation with DNA. Science, 1994,
266: 993－994
1 Watson J D, Hopkins N H, Roberts J W, et al. Molecular Biology
20 Gehani A, LaBean T H, Reif J H. DNA-based cryptography. Di-
of the Gene. 4th ed. Menlo Park, CA: The Benjamin/Cummings
macs Series In Discrete Mathematics & Theoretical Computer Sci-
Publishing Co., Inc., 1987
ence, 2000, 54: 233－249
2 Seeman N C. Nanotechnology and the double helix. Scientific 21 Celland C T, Risca V, Bancroft C. Hiding messages in DNA
American, 2004, 290: 34－43 microdots. Nature, 1999, 399: 533－534
3 Li Debao, Xu Ping. Theory and Methods of Recombinant DNA. 22 Leier A, Richter C, Banzhaf W, et al. Cryptography with DNA bi-
Hangzhou: Zhejiang Science and Technology Publishing Co., 1994 nary strands. Biosystems, 2000, 57: 13－22
4 Fodor S P, Read J L, Pirrung M C, et al. Light-directed, spatially 23 Shannon C E. Communication theory of secret systems. Bell Sys-
addressable parallel chemical synthesis. Science, 1991, 251: 767－ tem Technical Journal, 1949, 28(4): 656－349
773 24 Diffie W, Hellman M. New directions in cryptography. IEEE
5 Pease A C, Solas D, Sullivan E J, et al. Light-generated oligonu- Transaction on Information Theory, 1976, 22(6): 644－654
cleotide arrays for rapid DNA sequence analysis. Proc Natl Acad 25 Rivest R L, Shamir A, Adleman L. A method for obtaining digital
Sci USA, 1994, 91: 5022－5026 signatures and public-key. Cryptosystems Communications of the
6 Schena M, Shalon D, Ronald W, et al. Quantitative monitoring of ACM, 1978, 21(2): 120－126
gene expression patterns with a complementary DNA microarray. 26 EIGamal T. A public-key cryptosystem and a signature scheme
Science, 1995, 270: 467－470 based on discrete logarithms. IEEE Transaction on Information
7 Shalon D, Smith S J, Brown P O. A DNA microarray system for Theory, 1985, 31(4): 469－472
analyzing complex DNA samples using two-color fluorescent 27 National Institute of Standards and Technology, NIST FIPS PUB
probe hybridization. Genome Res, 1996, 6(7): 639－645 46-2, “Data Encryption Standards,” U.S. Department of Commerce,
8 Adleman L, Molecular computation of solutions to combinatorial 1993
problems. Science, 1994, 266: 1021－1023 28 Daemen J, Rijmen V. The Design of Rijndael: AES the Advanced
9 Guarnieri F, Fliss M, Bancroft C. Making DNA add. Science, 1996, Encryption Stand. Berlin: Springer-Verlag, 2002
273: 220－223 29 Shor P W, Algorithms for quantum computation: Discrete log and
10 Bancroft C, Bowler T, Bloom B, et al. Long-Term storage of in- factoring. Proceedings of the 35th Symposium on Foundations of
formation in DNA. Science, 2001, 293: 1763－1765 Computer Science. Los Alamitos, CA: IEEE Computer Society
11 Ouyang Q, Kaplan P D, Liu S, et al. DNA solution of the maximal Press, 1994. 124－134
clique problem. Science, 1997, 278: 446－449 30 Grovel L K. Quantum mechanics algorithm for database search. In:
12 Sakamoto K, Gouzu H., Komiya K, et al. Molecular computation Proceedings of the 28th ACM Symposium on the Theory of Com-
by DNA hairpin formation. Science, 2000, 288: 1223－1226 putation. New York: ACM Press, 1996. 212－219
13 Ravinderjit S, Braich R, Chelyapov N, et al. Solution of a 31 Simon D. On the power of quantum computation. In: Proceedings
20-Variable 3-SAT problem on a DNA Computer. Science, 2002, of the 35th Symposium on Foundations of Computer Science. Los
266: 499－502 Alamitos, CA: IEEE Computer Society Press, 1994. 116－123
14 Fastest DNA Computer. Science, 2005, 308: 195 32 Wiesner S. Conjugate coding. SIGACT News, 1983, 15: 78－88
15 Liu Q, Wang L, Frutos A G, et al. DNA computing on surfaces. 33 Bennett C H, Brassard G. Quantum cryptography: Public-key dis-
Nature, 2000, 403: 175－179 tribution and coin tossing. In: Proceedings of IEEE International
16 Roweis S, Winfree1 E, Burgoyne R, et al. A sticker based model Conference on Computers, Systems and Signal Processing. India:
for DNA computation. Journal of Computational Biology, 1998, Bangalore Press, 1984. 175－179
5(4): 615－629 34 Bennett C H. Quantum cryptography using any two nonothogonal
17 Lipton R J. Using DNA to solve NP-complete problems. Science, states. Phys Rev Lett, 1992, 68(21): 3121－3124
1995, 268: 542－545 35 Ekert A K. Quantum cryptography based on Bell's theorem. Phys
18 Adleman L M, Rothemund P W K, Roweiss S, et al. On applying Rev Lett, 1991, 67(6): 661－663
molecular computation to the Data Encryption Standard. Journal of 36 Bennett C H, Brassard G, Ekert A K. Quantum cryptography. Sci-
Computational Biology, 1999, 6(1): 53－63 entific American, 1992, 267: 50－57

1420 Chinese Science Bulletin Vol. 51 No. 12 June 2006