Learning-based Genome Codec


Harnessing the power of neural network techniques, this computational codec (encoder–decoder) compresses genome sequence data files with much higher compression ratio than traditional tools, and is now the core technique in a new standard for genome compression.

  • Learning-based Genome Codec 0
  • Learning-based Genome Codec 1
Research completion
Commercialisation opportunities
Technology licensing agreement
Problem addressed

Relying on artificial intelligence, this invention focuses on high efficiency genome compression through machine learning and context modeling. The prototype software can achieve more than 100 compression ratio over test data, featured with configurable, low-complexity, and high compression efficiency for genome data compression. As such, storage and transmission cost can be effectively reduced for genome data.

  • By using machine learning models, we can compress the genome data of the same specie with high efficiency.
  • In order to improve the compression speed, we innovatively introduced three prediction methods, including parallel prediction, multi-stride prediction and bidirectional prediction.
  • These predictive models can effectively reduce the computational complexity of machine learning models during compression.
Key impact
  • Efficient: The compression ratio of our invention achieves more than 100 times, which is much higher than traditional encoder (e.g., 4x)
  • Fast: Compared with methods without codec optimization, our technologies boost the processing speeding (50x speed up), facilitating a variety of applications.
  • Intelligent: With the compact representation, we can explore the relationship of different species, providing the fundamental support for biology research.
  • 48th International Exhibition of Inventions Geneva (IEIG) - Silver Medal
  • Genetic cloud database
  • Telemedicine
  • Genetic diagnosis


  • Patent Priority No. 17/445,202 (USA) IDF 996
City University of Hong Kong (CityU)

As one of the fastest growing universities in the world over the past decade, City University of Hong Kong (CityU) is recognised as a hub for innovation in research and professional education. CityU identifies solutions to critical global challenges by extending the frontiers of knowledge both within and beyond existing research paradigms.

The University’s highly qualified academics are drawn from all over the world, not only bringing a wealth of research and professional experience to the teaching programmes, but also contributing to the knowledge and technology advancement.