How Alibaba’s AI‑Driven Genome Platform Slashes COVID‑19 Test Time to 30 Minutes

Alibaba DAMO Academy is applying AI algorithms to combat the COVID‑19 pandemic.

On February 1, the Zhejiang Provincial CDC launched an automated whole‑genome detection and analysis platform. Using AI models developed by DAMO Academy, the time required for suspect case genome analysis was reduced from several hours to about half an hour, enabling rapid and precise detection of viral mutations.

Rapid, accurate diagnosis remains critical as the virus, with a genome length of 29,847 bp, requires full‑genome comparison of patient samples for confirmation. Conventional nucleic‑acid tests only target partial regions and may miss mutated strains.

The new platform, jointly created by the Zhejiang CDC, Alibaba DAMO Academy, and Jieyi Biotechnology, employs high‑throughput sequencing and fully automated library preparation, cutting manual preparation from 12 hours to 2 hours. Distributed algorithms accelerate sample analysis to 30 minutes and virus assembly to 15‑30 minutes, while providing complete genome coverage to avoid missed detections.

During the gene‑analysis stage, the system offers fast virus assembly, capturing mutated sequences and predicting secondary and three‑dimensional structures, supporting vaccine and drug development.

According to Dr Sun Yi, head of gene sequencing at the Zhejiang CDC, the platform’s cloud‑based computing power and DAMO’s novel algorithms can quickly scale to cover all confirmed cases and lay a solid foundation for future vaccine and therapeutic research.

Q: What are the platform’s key features?

A: It is a high‑throughput sequencing platform that provides fully automated library construction and distributed computational analysis. Jieyi’s instrument reduces library preparation time, while DAMO’s distributed algorithms cut analysis time from hours to half an hour and virus assembly time from up to an hour to 15‑30 minutes. It also delivers full‑genome detection, preventing missed cases caused by viral mutations.

Q: What value do DAMO’s algorithms bring?

A: The team analyzed coronavirus genomic features and trained models on public datasets such as PDB. Open‑source algorithms were optimized with distributed designs, and a BiLSTM + DNN model predicts protein secondary structure. Ongoing research includes 3‑D structure prediction and drug‑screening models.