Professor Ye Kai's team from the Ministry of Education's Key Laboratory for Intelligent Networks & Networks Security at Xi'an Jiaotong University (XJTU) published a research paper in the journal Cell titled The evolution of high-order genome architecture revealed from 1,000 species.

For the first time, from the perspective of "system architecture", the study reveals how the evolution of 3D higher-order genomic structures laid the physical foundation for the emergence of biological complexity.

During the research, the team collaborated with Professor Meng Deyu's team from Macau University of Science and Technology to construct the underlying mathematical models. Through long-term cooperation, the two teams have formed a close-knit research group, pioneering a "knowledge-data dual-driven" research paradigm for life sciences.

By analyzing 3D genomic data from 1,025 species (including bacteria, fungi, plants, and animals) spanning 3.8 billion years, they systematically mapped the evolutionary panorama of higher-order genomic structures.

While "structure determines function" is a cornerstone of biology, how the 3D structure of the genome itself evolved to support the transition from simple to complex systems remained a mystery. To extract universal laws from massive, heterogeneous cross-species data, the team developed an original method merging automated science with artificial intelligence (AI).

Professor Ye's team proposed converting linear gene sequences or complex 3D interaction data into feature images recognizable by computer vision, capturing deep patterns within life's blueprints.

To handle noise and heterogeneity in evolutionary data, Professor Meng's team built a model centered on sparse representation and bilevel optimization algorithms. This system acts as a high-precision "denoising" tool, extracting true evolutionary signals from chaotic data.

The research explicitly defines and quantifies two core types of higher-order genomic architecture:

1. Global folding

This describes the overall spatial arrangement of chromosomes within the nucleus, much like the load-bearing structure of a building. While universal across life, its strength does not correlate with biological complexity. Plants, however, have significantly reinforced this structure, likely to cope with environmental stresses linked to their sessile (immobile) nature.

2. Checkerboard

This reflects the degree of spatial separation between active and repressed genomic regions, similar to urban functional zoning. Crucially, the strength of the checkerboard pattern is positively correlated with biological complexity. More complex organisms move toward more refined departmentalized management of their 3D genome.

The analysis reveals a fundamental divergence in evolutionary strategies. Animals weakened the rigid constraints of global folding to make room for the checkerboard – the "software architecture" required for highly differentiated cell types and complex behaviors.

Plants leaned into reinforcing the "hardware skeleton" of global folding, relying on other regulatory methods (like linear gene clusters) for adaptation.

This macro-evolutionary discovery finds a parallel in human embryonic development. The study shows that early human embryos undergo a transition from strong global folding to a strong checkerboard, synchronizing with the shift from totipotency to specialized cell function. This suggests that the transition from a "stable skeleton" to "dynamic zoning" is a universal law of life across different timescales, from billions of years of evolution to mere days of development.

This study unveils the 3D genomic laws driving the evolution of life's complexity, and demonstrates a cutting-edge interdisciplinary paradigm: treating living organisms as dynamic, self-organizing complex systems and using systems modeling and AI to reverse-engineer their core architecture.