CAREER: Integrative modeling of intercellular interactions in the tumor microenvironment

乳腺癌是女性癌症死亡的第二大原因，也是美国非裔美国人和西班牙裔女性癌症死亡的主要原因。该项目的目标是研究乳腺癌肿瘤内部和周围细胞之间的相互作用，以更好地了解侵袭性肿瘤如何逃避人体的免疫防御。将开发新的计算方法来分析免疫细胞和乳腺癌之间的相互作用。这个新的框架将揭示乳腺肿瘤中的细胞组织如何影响免疫反应，并将为提高抗肿瘤免疫力的方法提供信息。开发的工具将公开提供，以使更广泛的科学界受益，用于探索其他生物系统和癌症。该研究计划将与弥合计算科学和生物医学工程之间差距的教育使命紧密结合。代表性不足的高中和本科生将通过与AI 4ALL和Hypothekids Bioforce计划的合作伙伴关系参与其中。研究生还将通过机器学习的新课程学习分析高维数据的计算技术。该CAREER项目将以深度生成框架的形式开发新型机器学习方法，用于分析高分辨率和高维空间转录组数据，以揭示乳腺肿瘤微环境中异质和相互作用的细胞状态。可扩展模型将为不同细胞状态之间的相互作用提供可解释性，特别是肿瘤相关成纤维细胞和免疫细胞之间的相互作用。将构建半监督模型，用于通过结合先前的生物学知识从空间转录组学数据推断细胞状态的组成。然后，该模型将被扩展到系统地推断细胞-细胞相互作用网络，通过整合转录数据与三阴性乳腺癌的侵袭性亚型的组织学成像。此外，将开发多变量模型，以通过整合单细胞表观遗传数据来学习基因-基因调控相互作用和空间动态免疫细胞群体的主要调控因子。该项目的发现将促进我们对免疫逃避机制的理解，并将指导调节乳腺癌免疫反应的策略。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Breast cancer is the second leading cause of cancer deaths among women overall and the leading cause of cancer deaths among African American and Hispanic women in the United States. The goal of this project is to study interactions between cells inside and around breast cancer tumors to better understand how aggressive tumors evade the body’s immune defenses. New computational methods will be developed to analyze the interactions between immune cells and breast cancer. This new framework will reveal how cell organization in breast tumors impacts immune response and will inform approaches for improving anti-tumor immunity. The developed tools will be made publicly available to benefit the broader scientific community for use in exploring other biological systems and cancers. The research program will be closely coupled with an educational mission of bridging the gap between computational sciences and biomedical engineering. Underrepresented high school and undergraduate students will be engaged through partnerships with the AI4ALL and Hypothekids Bioforce programs. Graduate students will also learn about computational techniques for analyzing high-dimensional data through a new course in machine learning. This CAREER project will develop novel machine learning methods in the form of deep generative frameworks for analyzing high-resolution and high-dimensional spatial transcriptomic data to uncover heterogeneous and interacting cell states in the context of the breast tumor microenvironment. The scalable models will provide interpretability for interactions between diverse cell states, in particular between tumor-associated fibroblasts and immune cells. Semi-supervised models will be constructed for inferring the composition of cell states from spatial transcriptomic data through incorporating prior biological knowledge. This model will then be extended to systematically infer cell-cell interaction networks by integrating transcriptional data with histology imaging of the tissue in the aggressive subtype of triple-negative breast cancer. Additionally, multivariate models will be developed to learn gene-gene regulatory interactions and master regulators underlying spatially dynamic immune cell populations by integrating single-cell epigenetic data. The findings of this project will advance our understanding of immune evasion mechanisms and will guide strategies for modulating immune response to breast cancer.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.