How Can DNA Computers Solve Specific Problems?
DNA is present in many living beings including humans. It is composed of nucleotides joined together to form strands. A nucleotide consists of a phosphate group, sugar and a nitrogenous base i.e. A(Adenine), T(Thymine), C(Cytosine) or G(Guanine). DNA is composed of 2 strands that bind together in a helix. In this binding only A-T and C-G combinations called base pairs can happen and that makes 1 strand complimentary to another. However, it is synthetic DNA that is being researched for computing power. In fact ability to create tailor made DNA is an important input to development of DNA computers.
Working
In 1994, Leonard Adleman at University of California, demonstrated a solution of 7 point directed Hamiltonian Path (HP) problem using DNA. This is closely related to more widely know Travelling Salesman Problem (TSP) where one has to find the shortest path through all cities visiting each city exactly once when path lengths between different cities are known. Leonard mapped each city as well as the paths to DNA strands and mixed the strands in a test tube. Various combination of strands emerged with road strand joining strands representing cities due to base pair binding. This happened in a second with each combination representing a possible path. Now Leonard had to work for many days to eliminate the incorrect ones. Multiple eliminations e.g. path not starting and ending on correct cities, each city is represented and represented only once etc was done using chemical reactions to get the correct answer. This was first proof of concept of a DNA computer. A DNA computer will not have a traditional memory, CPU or hard disk.
In 2000, experiments at Wisconsin University fixed DNA strands representing all possible solutions to a problem to a gold plate. The strands were exposed to complimentary strands representing logical conditions and resulted in forming of helixes. Enzymes were used to eliminate left over single strands. Then the joined ones were heated to give off their complimentary strands. The process was repeated multiple times to successfully impose different logical conditions. This demonstrated a more practical DNA computer.
Logic gates (AND, OR, NOT) have been constructed for DNA computer. In one method of strand replacement, a DNA strand joins a gate and ejects a strand that acts as output that can serve as input to the next gate. Gates can also be constructed using enzymes that can be used to ligate (join) and cut strands. Microsoft has actually created a programming language, DNA Strand Displacement (DSD) tool. It is a based on DNA strand replacement method and is used to design devices solely in terms of nucleic acids.
It is possible to have a inter woven sheets of DNA strands called tiles to self assemble into larger structures. This process of self assembly could be mapped to solve a computation problem. Similarly, single strands of DNA could undergo folding multiple times to become 2 dimensional structures called Origami. They are mostly used as nanorobots but the interactions of nanorobots and controls can be mapped to computer program. Most of the DNA circuits are built for 1 set of inputs. Renewable or reversible DNA computing that can work on multiple inputs is an area of active research.
Advantages and disadvantages
DNA offers incredible densities. Assuming a nucleotide can be stored in 1 square nanometer, it can store 1 million Gbits of information per square inch which is 100,000 times better than that of a hard disk.10 trillion DNA molecules fit in single cubic centimeter of volume and this can create a parallelism of computing power of similar magnitude. Also, this can be increased just by increasing DNA whose supply is not an issue. DNA is stable under moderate weather conditions. Scientists have sequenced or read DNA from a 430,000-year-old Neanderthal man and 500,000-year-old horse. It does not need any significant energy in processing. This can be compared to supercomputers who consume power in MWs.
However, advantages of DNA are also its inherent disadvantages. As a DNA computer creates all possible solution whether correct or incorrect, it needs space for that many strands sometimes called “memory”. At the current level of technologies, TSP of even a few hundred cities may need DNA comparable to the weight of the Earth. This effectively shifts the problem of exponential time resources to exponential space resources. The elimination of unwanted strands to get the correct answer is a time-consuming process requiring human intervention. All parallel processing methods e.g. Supercomputers, Quantum computers etc need efforts to reconcile the results after processing but it is particularly resource intensive and manual in DNA computers. Also, as strands may break or join incorrectly, or enzymes may cut or insert incorrectly, errors are common and they increase significantly as number of strands increase. DNA can be damaged by heat or Ultraviolet radiation from Sun.
At the current state of development, DNA computers cannot solve problems that cannot be solved in polynomial time on conventional computers.
Storage
There has been significant progress in the related field of using DNA for storage. In 2012 Harvard University researchers encoded a 52,000-word book. But in 2017 scientists at Columbia University and the New York Genome Center demonstrated storage and retrieval of a full computer operating system, an film gift card, a computer virus, plaque etc at densities of 215 petabytes per gram of DNA. Microsoft has plans to use DNA for data storage in its cloud.
Conclusion
DNA could be used as storage medium. It can also be used inside living organisms to detect diseases or treatment. However, when it comes to its being used as a computer, it will not replace the conventional computers due to its multiple shortcomings e.g. time needed to remove incorrect answers and efforts required to reduce errors. Also making long strands that can encode more information has cost implications. DNA computers will be used to solve specific problems where solutions can be worked in parallel e.g. cryptography, scheduling problems, finding efficient routes in transportation etc. They will also be suitable for “fuzzy logic” problems that have many possible solutions. When DNA computers arrive, they will complement the conventional computers for solving such problems.
