Keren Group

Decoding Cancer Ecosystems 

Tumors are complex ecosystems in which malignant cells interact with immune, stromal, and vascular cells to shape cancer progression and response to treatment. Our goal is to understand how these interactions govern tumor behavior and translate this knowledge into better cancer diagnosis and therapy. To achieve this, we develop cutting-edge experimental and computational technologies that enable us to measure, model, and interpret the molecular and spatial organization of human tissues. By integrating spatial proteomics, artificial intelligence, and cancer biology, we aim to uncover the fundamental principles governing cellular interactions within tumors and advance the next generation of precision oncology.

Our research 

Next-Generation Spatial Proteomics

We develop technologies to measure proteins directly within intact human tissues at single-cell and subcellular resolution. Building on our pioneering work in multiplexed imaging, including MIBI-TOF and CombPlex, we aim to overcome current limitations in molecular coverage and throughput, ultimately enabling proteome-scale characterization of human tissues.

Artificial Intelligence and Computational Spatial Biology

We develop advanced computational methods to interpret high-dimensional tissue images, from cell segmentation and phenotyping to identifying multicellular neighborhoods and tissue organization. Our work includes widely adopted computational tools and next-generation AI foundation models that integrate spatial datasets across technologies, patients, and cancer types to uncover generalizable principles of tissue biology.

Deciphering the Tumor Microenvironment

We apply these technologies to large cohorts of human tumors to investigate how malignant, immune, and stromal cells interact to influence cancer progression, metastasis, and therapeutic response. By linking molecular and spatial organization to clinical outcomes, we aim to uncover mechanisms of tumor immunity and resistance, discover new biomarkers, and identify opportunities for therapeutic intervention.

Our vision

We envision a future in which the molecular and spatial architecture of every tumor can be comprehensively measured, understood, and used to guide clinical decisions. By combining proteome-scale spatial profiling with powerful AI models and insights into tumor biology, we aim to transform cancer pathology from descriptive tissue examination into a quantitative, predictive science, enabling earlier diagnosis, more accurate prediction of disease progression and treatment response, and therapies tailored to the unique biology of each patient’s tumor.