Seer, Inc. and Korea University presented preliminary data at the 2026 American Society for Mass Spectrometry Annual Conference showing the potential of AI-driven plasma proteomics for multi-cancer screening.
The findings came from more than 5,500 plasma samples across ten major cancer types and healthy controls. Sang-Won Lee, PhD, Professor at Korea University and Chief Executive Officer of TargetX, and Jaewoo Kang, PhD, Professor at Korea University and Chief Executive Officer of AIGEN Sciences, presented the data during the Seer Breakfast Symposium on June 1 in San Diego.
The ongoing collaboration between Seer and Korea University focuses on generating one of the largest deep, unbiased plasma proteomics datasets assembled to date. The researchers used Seer’s Proteograph® Product Suite and an Orbitrap Astral mass spectrometer to support deep plasma proteome profiling at population scale.
The researchers reported deep and reproducible plasma proteome coverage across the study cohort, with an average of more than 14,000 protein groups per sample. The work also supported construction of a comprehensive cohort-derived proteomic reference resource.
Working in analytical science?
Register for a FREE Separation Science account to subscribe to the Separation Science Newsletter.
“By incorporating the Proteograph platform into our plasma proteomics workflow, we are able to profile the plasma proteome at remarkable depth and scale across thousands of patient samples with the reproducibility required for large-scale clinical research,” Dr. Lee commented. “Equally important, the quality and consistency of the data enable us to explore new computational approaches that extend beyond conventional identification-based analyses. By combining deep proteomic datasets generated using the Proteograph and Orbitrap Astral mass spectrometer combined with our ID-Free AI framework, we can learn directly from a substantially larger portion of the underlying data and uncover biological patterns that may otherwise remain inaccessible.”
The researchers also described an emerging ID-Free AI framework designed to use the substantial portion of mass spectrometry data that conventional identification-based workflows leave uncharacterized. This approach creates a foundation for evaluating whether additional biological information within complex proteomic datasets can improve future multi-cancer screening strategies.
“The Proteograph platform significantly expands the breadth and depth of proteomic information captured from plasma samples,” Dr. Kang comments. “Applying self-supervised AI models directly to these rich datasets allows us to extract biological signal beyond what traditional workflows can utilize. This creates an opportunity to investigate whether previously untapped information can contribute to future multi-cancer screening approaches.”
The collaboration expects to analyze more than 20,000 clinical plasma samples across ten of Korea’s highest-incidence cancer types.
This article is based on a press release issued by Seer.
