Key points from article :
Ginkgo Bioworks has launched a major new effort to accelerate AI-driven drug discovery: the Virtual Cell Pharmacology Initiative (VCPI), led by John Androsavich, and presented through the company’s Ginkgo Datapoints platform. The initiative aims to create the largest open-source pharmacology dataset ever assembled—over 12 billion data points from tests on at least 100,000 compounds. By bringing together researchers, biotech companies, and AI developers, the project seeks to build a shared, standardized foundation for virtual cell modelling, a field that has long struggled with inconsistent data and fragmented methods.
Virtual cell models promise to simulate how drugs affect living cells, but their usefulness has been limited by two major bottlenecks: the lack of a common reference cell line and the scarcity of high-quality, reproducible pharmacology data. VCPI addresses both issues head-on. First, it introduces V-Ref293, a new engineered reference cell line that will be distributed globally from 2026, allowing labs worldwide to generate comparable results. Second, the initiative adopts DRUG-seq, a high-throughput bulk RNA sequencing technique validated for pharmaceutical use, offering a scalable and more reliable alternative to noisier single-cell approaches.
A distinctive feature of VCPI is its open, community-driven design. Instead of releasing fixed datasets, Ginkgo invites contributors to shape the project from the outset by submitting compounds for free high-throughput RNA profiling. Data will be released publicly under Creative Commons licensing, although contributors may choose temporary exclusivity or private use. The platform also includes voting mechanisms, model sharing, competitions, and a discussion forum, with the most active participants gaining “super user” status and early access to incoming data. Through this combination of scale, standardization, and openness, the VCPI aims to set a new benchmark for training high-quality AI systems in drug discovery.
