ثورة في اكتشاف الأدوية: كيف تختصر منصة GalaxyVS المدعومة بالذكاء الاصطناعي زمن الفحص إلى ثوانٍ معدودة

China has officially unveiled GalaxyVS, an artificial intelligence virtual screening platform that radically compresses the most time-consuming phase of early drug discovery. Developed by the National Supercomputing Center in Tianjin and Tsinghua University’s Institute for AI Industry Research, the system runs on the new-generation Tianhe supercomputer and can search a library of nearly 100 billion synthesizable chemical compounds in seconds . The breakthrough, first reported by Science and Technology Daily on May 27, 2026, marks what developers call the world’s first second-level virtual screening of a chemical space approaching 100 billion molecules .

What GalaxyVS actually is

GalaxyVS is not a single algorithm but a full-stack platform that marries AI-powered molecular modeling with extreme-scale parallel computing. The core research team is led by Meng Xiangfei, chief scientist at the National Supercomputing Center in Tianjin, and Professor Lan Yanyan of Tsinghua University. The platform was purpose-built to overcome the notorious bottleneck in early-stage drug discovery: finding a few promising lead molecules inside an effectively infinite sea of chemical possibilities .

Traditional high-throughput screening in physical labs handles roughly millions of compounds. GalaxyVS performs virtual screening across roughly 100 billion—an expansion that directly addresses what team member Dr. Li Peishun described as the trio of persistent problems: “sparse active molecules, insufficient searchable space, and homogeneity of candidate molecules” .

How a 100-billion-molecule screening finishes in seconds

The speed comes from architectural choices at every layer. The system first uses thousands of digital signal processing nodes to encode approximately 4 million target pockets spanning all model biological species. These encoded pockets are then searched across more than 20,000 compute nodes in parallel. A complete scan of the entire compound library finishes in tens of seconds. On a per-target basis, the average retrieval time drops below one second. The platform’s daily throughput reaches about 16 trillion molecular docking operations, which its developers report is six orders of magnitude higher than the previous international supercomputing record .

Put differently, what once required teams of medicinal chemists months or years to accomplish—the initial virtual screening of a massive compound library—now runs as a single job on the Tianhe system. This does not eliminate later stages like hit validation, lead optimization, or clinical trials, all of which remain lengthy and expensive. But it collapses the front-end search from a human-scale timeline into a computational one, freeing researchers to invest more time on the most promising candidates.

Diseases in the crosshairs

The team has outlined a wide therapeutic mandate. GalaxyVS is expected to accelerate lead molecule discovery for tumors, neurodegenerative conditions, rare diseases, and emerging infectious diseases . One of the most immediately cited use cases is responding to sudden public-health threats: the platform’s speed could compress the earliest research timeline during an outbreak, quickly identifying lead molecules that merit urgent wet-lab and preclinical follow-up .

This positioning is consistent with a broader Chinese research trend. Earlier work combining the Tianhe-2 supercomputer and AI algorithms had already demonstrated the feasibility of predicting biosynthetic pathways for natural products and engineering novel chemicals . GalaxyVS extends that lineage from pathway prediction to ultra-large-scale virtual screening, making the chemical search space practically searchable for the first time.

The official source and media corroboration

All the technical details and claims originate from the May 27, 2026, report in Science and Technology Daily (科技日报), the official newspaper of China’s Ministry of Science and Technology . The story was subsequently corroborated and expanded by English-language coverage in the South China Morning Post and Chinese-language outlets including Sina Finance and NetEase . No peer-reviewed paper detailing the GalaxyVS architecture had been published at the time of the announcement; the public information is based on official institutional statements and these media reports.

GalaxyVS is ultimately an engineering story as much as a scientific one. It shows what becomes possible when AI-driven molecular prediction is paired with machine scale that matches the size of the search space. The next steps—validation against real biological targets, integration into pharmaceutical pipelines, and independent reproduction by other labs—will determine how much of those tens of seconds turn into actual therapies.