A recent panel discussion at the 2024 Life Science Innovation Northwest meeting in Seattle focused on the impact of AI on biopharma and the new business models driven by AI. Panelists emphasized the culture of collaboration in the Pacific Northwest, highlighting the region’s natural hub for connecting AI with biotech. The Institute for Protein Design (IPD) at the University of Washington plays a central role in this collaboration, generating open-source AI tools for crafting protein-based therapeutics, vaccines, materials, and biosensors. The region’s spinouts and affiliated companies also partner with larger biopharma companies, with big tech companies like Microsoft investing in the area by releasing open-source models.
One major goal in the field of AI in biopharma is to not only discover new therapeutic proteins but also to shorten their clinical development through the concept of “quality by design.” Researchers can now assess proteins for specific traits like ease of manufacture or unwanted cross-reactivity to other molecules. Bristol Myers Squibb, a major player in Seattle, uses machine learning to mine internal data in order to match patients for clinical trials and conduct “virtual” clinical trials. The use of AI tools is enabling companies to accelerate drug development to the clinic faster and at a lower cost.
Advances in AI and biotech are focused on predicting the biological properties of computationally-designed therapeutics, with many of these developments originating from research conducted in Washington state. Efforts are also being made to standardize and share biological data to enhance the power of AI models. Pharma companies are investing in consortiums to share data and models, with Seattle-based consortium OpenFold leading the way. However, a challenge remains with high-quality data being locked up in proprietary databases, posing implications for how companies differentiate themselves in the industry.
Panelists discussed the concept of proprietary data as a differentiator for companies in a landscape of open-source AI models. They highlighted a potential business model where companies add their own data, fine-tuning, and internal expertise to open-source platform tools to create unique offerings. The notion of shared community foundation models being a platform for individual companies to build upon economically makes sense in the life sciences industry. It is emphasized that the origins of these platform models, like those from the IPD, being from publicly-funded universities adds a unique asset to the region and the world at large.
The balance between keeping data proprietary and sharing it for the greater good is still being worked out by researchers, institutions, and companies in the life sciences ecosystem. Data may become the commodity around which new companies are built, with implications for how companies approach their foundation models. The question of how much data should be shared versus kept proprietary is a crucial one for the community and ecosystem to address, as it shapes the future of the industry and the emergence of new companies.
