CRI Funded Scientists

Karin Pelka, PhD, Technology Impact Award Grantee

The J. David Gladstone Institutes

Area of Research: Colorectal Cancer, Cáncer de pulmón

The discovery of new immunotherapy targets for solid tumors is limited by our poor understanding of howvimmune, stromal, and malignant cells influence each other. The Pelka Lab recently identified several spatially-organized interaction networks between immune and malignant cells characterized by distinctive gene programs. These ‘immune hubs’ were conserved across patients and associated with immunotherapy response in colorectal
and lung cancer.

However, prioritizing candidate genes predicted to induce or remodel hubs is needed to experimentally test them individually and in combination as drug targets. Thus, the Theodoris Lab developed a foundational deep learning model, Geneformer, pretrained on about 30 million human single cell transcriptomes to gain a fundamental understanding of gene networks, identify their core elements, and nominate disease network-correcting therapies. Here, Dr. Pelka builds on Geneformer to develop computational models that identify central regulators that drive malignant cell state transitions and shape tumor immune landscapes.

Specifically, she aims to develop (1) a generative deep learning model to determine central regulators that induce the transition between immunologically distinct healthy, pre-malignant, and malignant epithelial cell states, and (2) a computer vision model for single cell-resolved spatial transcriptomic data sets, which her lab has already generated successfully, to determine how perturbations in the malignant cells impact the immune and stromal environment. She expects this work to identify new immunotherapy targets in colorectal cancer that, of note, are conserved across patients. Moreover, Dr. Pelka’s foundational models will be broadly applicable to many other biological settings including other human tumors, premalignant lesions, or immunotherapy-induced toxicities.

Projects and Grants

Foundational deep learning models to identify central regulators of malignant-immune interactions in human colorectal cancer

The J. David Gladstone Institutes | Colorectal Cancer, Lung Cancer | 2024

This website uses tracking technologies, such as cookies, to provide a better user experience. If you continue to use this site, then you acknowledge our use of tracking technologies. For additional information, review our Privacy Policy.