Читать книгу DNA- and RNA-Based Computing Systems - Группа авторов - Страница 17

Humans are looking for new approaches to computing since the starting of civilization. Over the years, researchers have invented many systems for computation, from “counting with abacus” to “complex computing by using modern‐day computers.” According to Moore's observation [1], the number of transistors on a silicon chip is found to be doubling in every 18–24 months, which results in the development of faster computing devices. However, in the coming decades, producing such faster computing devices will be more challenging as the size of the transistor is already approaching to a molecular level. Moreover, engineering such silicon chips is gradually becoming more complex and less cost effective. This compelled the researchers to look for alternative computing devices and methodologies. Biomolecular computing is one such excellent alternative to traditional silicon‐based computing methods.

Biomolecular computing is illustrated for the first time by Adleman in 1994 [2] to solve the Hamiltonian path problem (HPP) using deoxyribonucleic acid (DNA). In such DNA‐based computing (referred to as DNA computing), enzymatic reactions and manipulations such as addition, amplification, and cutting of the DNA are used for performing the computing. Subsequently, DNA computing is used by several researchers [3–8] to solve a variety of combinatorial problems. These studies have exploited high parallelism of DNA reactions over sequential operations occurring in silicon‐based computers to solve the computationally intractable problems. Such parallel processing in DNA computer builds the confidence for solving the problems that are presently not solvable with silicon‐based computers. Moreover, DNA became an effective and efficient material for faster computation, storage, and information processing owing to the significant advancements in biomolecular techniques such as gel electrophoresis, polymerase chain reaction (PCR), affinity separation, restriction enzyme digestion, etc. [9,10].

Despite initial success, the increase in problem size leads to a significant bottleneck in scaling the existing DNA computing procedures for large size problem formulations. This is primarily because the amount of DNA required increases exponentially with size even though the number of biochemical steps required increases with a polynomial function. Further, there is a significant compounding of experimental errors involved, which makes these procedures redundant for solving the bigger size formulations. Also, the real‐life problems often have continuous search spaces and multiple optimal solutions, whereas the existing DNA computing procedures are mostly developed for discrete search spaces involving a single optimal solution.