Figure: Workflow of general-purpose computational programming with DNA-based programmable gate arrays

With the support from the National Natural Science Foundation of China (Grant numbers T2188102, 21991134, 21904060, 22025404 and 22104088), a team led by Chunhai Fan and Fei Wang from School of Chemistry and Chemical Engineering, Frontiers Science Center for Transformative Molecules, Shanghai Jiao Tong University developed DNA-based programmable gate arrays (DPGAs) that allows general-purpose digital computing. Using DPGAs, they demonstrated general-purpose digital DNA computing through programming with molecular instructions and achieved the construction of large-scale liquid-phase molecular circuits without attenuation. The research results were published in Nature on September 13, 2023, with the title “DNA-based programmable gate arrays for general-purpose DNA computing” (link at https://www.nature.com/articles/s41586-023-06484-9).

In 1994, Turing Award winner Adleman proposed using the base-pairing principle of DNA to develop bio-computation. Since then, liquid-phase DNA molecular computing based on intermolecular interactions of DNA has shown tremendous potential in high parallel encoding and algorithm execution. Researchers have successfully utilized DNA molecular reaction networks to realize various functions, such as cellular automata, logic circuits, decision making machines, and neural networks. However, existing DNA computing systems are mainly limited to hardware customization for specific functions. In the field of electronic computing, general-purpose integrated circuits (e.g., FPGA) can execute various computational functions through software programming without de novo designing and manufacturing hardware, providing a higher-level platform for developing computing devices. Therefore, how to develop general-purpose DNA computing units and realize their programming and integration has become a bottleneck restricting the development of the field of DNA computing.

To address this challenge, the research team first demonstrated that using single-stranded DNA as a uniform transmission signal (DNA–UTS) can achieve signal transmission similar to that of electrons in electronic circuits. They further developed a DPGA that supports general-purpose digital computing and established a method for multi-DPGA integration at the device level, achieving intra-device programmability and inter-device integrability. When the complexity of a circuit exceeds the executable scale of a single DPGA, the circuit can be divided into subcircuits and corresponding molecular instructions can be generated. The molecular instructions for each sub-circuit call and connect the participating DNA units through their logical addresses, realizing the programming of DPGA. Signal transmission between sub-circuits and multi-DPGA routing is mediated by DNA origami registers, thereby achieving device-level multi-DPGA integration.

Leveraging the programmability and high integration of DPGAs, this study broke through the bottleneck in circuit scale and circuit depth of DNA molecular computing. For the first time, they experimentally demonstrated a circuit scale with up to 30 logic gates, 500 DNA strands, and 30 layers of DNA strand displacement reactions, representing a new breakthrough in the field of DNA computing. The research found that the DNA-UTS did not significantly attenuate when transmitted across multiple DPGAs, proving the high scalability of DPGAs. Theoretically, it ensures that any practical problem can be connected into DPGA circuits via DNA-UTS after analog-to-digital conversion. This study conceptually demonstrated nonlinear classification of disease-related molecular targets using DPGA as the information processing core in molecular diagnostics. The ability of DPGA to integrate and operate large-scale reaction networks signifies a crucial step toward general-purpose DNA computing, and it holds promise for broader applications in mathematical computation and disease diagnosis.