Digital life, as a cutting-edge interdisciplinary field integrating life sciences and digital technologies, has become a pivotal research direction driving the paradigm shift of life science research. This paper aims to systematically trace the evolutionary trajectory of digital life from theoretical hypothesis to engineering implementation, deeply dissect its core technical system and current development status, identify the key bottlenecks restricting its in-depth development, and propose targeted development strategies and future prospects, thereby providing a comprehensive reference for academic research, industrial transformation and the construction of governance systems in this field. The study first combs the multi-stage development process of digital life: starting with von Neumann′s self-replicating automaton theory that laid the theoretical foundation of artificial life, it has experienced the stage of large-scale accumulation of life omics data driven by high-throughput sequencing and super-resolution imaging technologies, the stage of breakthroughs in biological feature decoding relying on synthetic biology and artificial intelligence (AI) models such as AlphaFold, and then entered the stage of practical application of digital twin technology, with the birth of a series of whole-organism models including AI virtual cells, digital organs, intelligent nematode ″Tianbao 1.0″, virtual rats and digital fruit flies. On this basis, the paper elaborates the four core technical pillars supporting the development of digital life: life data acquisition with high-throughput, ultra-high resolution and multi-modal characteristics, life feature decoding empowered by large models such as transformers and graph neural networks, life activity simulation emphasizing wet-dry closed-loop validation and multi-scale model iteration, and the integrated life digital twin system that realizes dynamic coupling between virtual and real entities. It also reveals the core connotation of digital life research in realizing the in-depth integration of spatial, temporal and logical dimensions, and constructing a digital system that conforms to the laws of real life and has controllability. In the research, it is found that digital life is currently in a critical period from concept verification to system engineering, and is facing prominent bottlenecks such as difficulty in integrating multi-source heterogeneous data, weak interpretability of cross-scale models, insufficient understanding of the dynamic causality of life systems, lagging ethical governance systems, and barriers in interdisciplinary collaborative innovation. In response to these problems, this paper holds that the future development of digital life should follow the progressive evolution path of ″division-conjunction-integration″. It is imperative to further deepen the theoretical understanding of the emergent mechanisms of life systems, develop new AI architectures embedded with biophysical constraints to improve the interpretability and predictability of models, establish standardized data platforms and model interfaces to break interdisciplinary barriers and promote open-source collaboration, and construct a comprehensive governance system covering intellectual property protection, ethical review and safety supervision. Only by overcoming the above technical and institutional bottlenecks can we truly realize the comprehensive parsing and reconstruction of the life panorama, promote the mature application of digital life in precision medicine, drug research and development, molecular engineering and other fields, and make digital life an important scientific infrastructure for revealing the mysteries of life and safeguarding human health.
ZHANG et al. (Wed,) studied this question.