Tempus and Genialis Partner on RNA-Based Biomarker Algorithms

Tempus AI has partnered with Genialis to develop RNA-based biomarker algorithms aimed at improving cancer diagnostics and treatment outcomes. This collaboration allows Genialis to use Tempus’ multimodal dataset, one of the largest collections of real-world patient data, to refine and validate its biomarker algorithms across various cancer types.

At the core of Genialis’ approach is its Genialis Supermodel, a foundational molecular model trained on approximately 1 million RNA sequencing samples. This model integrates data from preclinical studies, single-cell analyses, and globally diverse patient records to create a high-dimensional landscape of cancer biology. The Supermodel is designed to transform raw RNA data into biologically relevant insights, eliminating systematic biases and condensing thousands of input features, such as gene expression and gene variants, into actionable biological states.

Credit: Genialis

The Genialis Supermodel processes each new patient sample by placing it within a detailed framework of cancer biology, derived from diverse RNA sequencing data. This framework enables the identification of biological states and processes relevant to the sample, supporting the generation of predictive biomarkers. These biomarkers stratify responders from non-responders, uncover mechanisms of therapeutic response or resistance, and propose potential combination therapies or novel drug targets.

Tempus’ multimodal dataset, accessed through its Lens analytics platform, would complement the Supermodel by providing the real-world data needed to validate these biomarkers in clinical settings. This collaboration aims to refine patient stratification and optimize drug development strategies, offering data-driven insights for more precise and effective cancer treatments.

See also: 19 Companies Pioneering AI Foundation Models in Pharma and Biotech

One achievement from this collaboration is the further development of algorithms like krasID, which stratifies patients likely to respond to KRAS-targeted therapies. Presented at the 2024 Targeting-RAS Drug Development Summit, krasID uses real-world data to predict patient responses across cancer types and KRAS mutations, marking a step forward in precision oncology.

Under the agreement, Tempus has the option to evaluate and potentially license biomarkers developed by Genialis for integration into its xR platform, which aims to enhance cancer care through data-driven tools. Both companies see this partnership as a way to address current limitations in cancer diagnostics by creating more accurate and clinically actionable biomarkers. While Genialis benefits from Tempus’ vast dataset and analytics capabilities, Tempus gains access to biomarker algorithms to expand its precision medicine offerings.

Topics: Startups & Deals