insitro, a machine learning-driven drug discovery and development company, and Genomics England today announced a collaboration to bring insitro’s machine learning powered embedding search capabilities to Genomic England’s multimodal phenotypic and genetic database. Since it was established in 2013 to deliver the 100,000 Genomes Project, Genomics England has built an unparalleled resource of almost 150,000 whole genomes with corresponding phenotypic data from NHS patients with rare diseases and their families, as well as patients with cancer. Included in this resource, Genomics England has a growing corpus of multimodal and high-dimensional data, including genetic data and histopathology images from its cohort of cancer patients.
As part of the collaboration, insitro will deploy its machine-learning capabilities on Genomics England’s histopathology images and accompanying genetic data to learn a multimodal representation (an embedding) that captures the fine grained semantic structure in this high-dimensional clinical data. Powered by this representation, insitro’s embedding search engine enables users to search for related images, biopsies, or cases based upon biologically and clinically relevant semantic similarity rather than visual similarity. Under the terms of the agreement, insitro will make its embedding search engine available to Genomics England’s network of research partners within the secure Genomics England Research Environment. This capability will enable Genomics England’s research partners to more fully realize the promise of this multimodal data resource, accelerating cancer research by empowering scientists to derive multimodal insights that go beyond traditional diagnostic labels.
insitro will also become a Genomics England research partner more broadly, partnering with Genomics England to empower insitro’s mission of bringing medicines to patients in need through the combination of genetic data and machine learning.
“We are pleased to partner with Genomics England to support their mission of deriving novel insights from their world-class collection of complex, multimodal data, which can help in providing better care and new therapeutic interventions to patients with rare diseases and cancer,” said Daphne Koller, Ph.D., founder and chief executive officer of insitro. “We believe that cutting edge machine learning methods can go beyond automating predefined prediction tasks and enable clinicians and scientists to derive novel insights that might give rise to new discoveries and treatments. Following on our open sourcing of our redun workflow engine in November of 2021, the development of our embedding search engine for Genomics England demonstrates insitro’s commitment to building a truly enabling data driven discovery ecosystem for the benefit of patients. We believe this collaboration can be an exemplar of future collaborations that unlock the power of high-content data towards improved therapeutics and outcomes.”
“As we continue to advance our research strategy of making genomic data available and usable to global biopharma and academic researchers, we are now moving towards a future of enabling deeper genomic research through multimodal data, bringing genomic healthcare to all who need it,” said Parker Moss, Chief Commercial Officer at Genomics England. “insitro’s expertise in extracting insights from multimodal clinical data with machine learning is unparalleled, and this partnership is a key step forward to help our research partners maximize discovery in our research environment and enable our shared vision for the future.”
For more such updates and perspectives around Digital Innovation, IoT, Data Infrastructure, AI & Cybersecurity, go to AI-Techpark.com.
