An ambitious project is underway globally to sequence the genomes of 1.85 million eukaryotic species on Earth. This work will greatly enhance human understanding of biology and provide key data support for species protection and biodiversity restoration.
In today's genome sequencing process, AI tools are widely used to collect, assemble and calibrate DNA data with high precision, ensuring that errors are minimized across billions of genetic fragments. The Earth BioGenome Project (EBP), which is jointly participated by more than 60 projects around the world, was launched in 2018 and is expected to cost nearly US$5 billion. So far, it has completed data contributions for 4,386 species, covering mammals, reptiles, fish, birds, insects, plants and other organisms.
The project is significant because, on average, dozens of species become extinct every day. Once they disappear, humanity will lose its last chance to understand how they evolved and survived. Comprehensive establishment of a genome catalog of approximately 2 million species will leave valuable resources for biological research and earth conservation.
Although EBP is currently only a small step away from the ambitious goal of sequencing all species, the rapid development of AI and automation technology in recent years is expected to accelerate the sequencing process. The Google research team plays an important role in this: its AI tool "DeepVariant" released in 2018 can accurately reconstruct an individual's complete genome using data generated by modern high-throughput sequencers (HTS). Since HTS instruments only generate billions of short fragments at a time, AI assembly of fragments into reliable complete sequences is a major challenge for sequencing. DeepVariant transforms this process into an image classification problem. The deep neural network analyzes the compared visual data to determine the authenticity of genetic mutations or instrument errors. The accuracy is significantly higher than traditional methods.
The Google research team also launched the "DeepPolisher" tool this year, which can reduce the error rate in the genome assembly process by 50% and prevent genetic variations that may cause disease from being missed during subsequent gene annotation. With tools like these, Google hopes to help researchers understand how species avoid disease risks and enable targeted interventions to protect endangered species.
The report takes the Kākāpō, an endangered parrot unique to New Zealand, as an example to illustrate the significance of genome sequencing for species conservation. In 2015, there were only 49 individuals of this species left. Scientists used the genetic data of the group to analyze its genetic diversity, migrate and breed individuals rationally, effectively avoiding the problems of low fertility and weak immunity caused by inbreeding. After three generations, the number of individuals has expanded to nearly 200. There are currently more than 250 parrots in the New Zealand Islands.
With the continuous advancement of AI and automation, the efficiency of the EBP team in completing sequencing work will also be greatly improved in the future. Compared with the first human genome sequencing, which took 10 years and cost up to US$3 billion, the process now only takes one day and the technical cost has dropped to US$1,000.
For more project details, please visit the official website of the "Earth Biogenome Project":
https://www.earthbiogenome.org/