PO.BCS01.09 · 生物信息与计算
STniche：一种从空间转录组学识别肿瘤微环境中功能性微环境的方法
STniche: An approach to identify functional niches in the tumor microenvironment from spatial transcriptomics
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
背景：空间转录组学保留了肿瘤微环境（TME）中的空间背景，能够精确绘制细胞相互作用和免疫结构。然而，将这些数据转化为功能性微环境图谱仍然困难。我们开发了STniche，这是一个整合框架，可识别空间连贯且具有生物学可解释性的功能性微环境，将局部转录组活性与组织构架联系起来。
方法：STniche使用通路特征为spot或细胞分配功能表型，通过局部Moran's I纳入相邻细胞间的空间依赖性，并借助基于高斯模型的聚类检测空间微环境。最优微环境数量通过BIC选择。我们将STniche应用于Meylan等人（2022）的透明细胞肾细胞癌（ccRCC）数据集，该数据集包含病理标注的三级淋巴结构（TLS）。在十个肿瘤切片中，我们使用一个整合了五个协调程序（趋化因子信号、淋巴毒素轴、抗原呈递、生发中心B细胞活化和树突状细胞成熟）的精选TLS元通路，以捕获TLS生物学的核心分子过程。将STniche识别的TLS微环境与病理标注进行比较。
结果：STniche能够识别与TLS对应的空间连贯的功能性免疫微环境。TLS元通路定义的聚类与病理学家标注高度一致，平均一致性为0.39（范围0.28-0.56），平均相对对称性为0.64（范围0.36-0.96）。平均而言，约42%的STniche定义spot与病理确认的TLS重叠，且这些聚类捕获了约42%的所有TLS阳性spot。这些发现表明，STniche能直接从空间转录组数据可靠地恢复功能性TLS微环境，为免疫微环境发现提供了定量且无监督的框架。
结论：STniche提供了一个统计学上严谨且具有生物学可解释性的框架，用于发现肿瘤组织中的功能性空间微环境。其检测TLS富集免疫结构的能力凸显了其在理解预后、免疫生物学和治疗反应方面的潜在应用。进一步的基准测试正在进行中。
查看英文原文 English abstract
Background: Spatial transcriptomics preserves spatial context in the tumor microenvironment (TME), enabling precise mapping of cellular interactions and immune architecture. However, translating these data into functional niche maps remains difficult. We developed STniche, an integrated framework that identifies spatially coherent and biologically interpretable functional niches that link local transcriptomic activity to tissue organization.
Methods: STniche assigns functional phenotypes to spots or cells using pathway signatures, incorporates spatial dependency between neighboring cells with local Moran's I, and detects spatial niches via Gaussian model-based clustering. The optimal number of niches is selected by BIC. We applied STniche to the Meylan et al. (2022) clear cell renal cell carcinoma (ccRCC) dataset containing pathology-annotated tertiary lymphoid structures (TLS). Across ten tumor sections, we used a curated TLS meta-pathway integrating five coordinated programs-chemokine signaling, lymphotoxin axis, antigen presentation, germinal-center B-cell activation, and dendritic-cell maturation-to capture core molecular processes of TLS biology. STniche-identified TLS niches were compared with pathology annotations.
Results: STniche was able to identify spatially coherent functional immune niches corresponding to TLS. TLS meta-pathway-defined clusters showed strong agreement with pathologist annotations, with mean concordance of 0.39 (range 0.28-0.56) and mean relative symmetry of 0.64 (range 0.36-0.96). On average, ~42% of STniche-defined spots overlapped pathology-confirmed TLS, and the clusters captured ~42% of all TLS-positive spots. These findings demonstrate that STniche reliably recovers functional TLS microenvironments directly from spatial transcriptomic data, providing a quantitative and unsupervised framework for immune niche discovery
Conclusion: STniche provides a statistically rigorous and biologically interpretable framework for discovering functional spatial niches in tumor tissues. Its ability to detect TLS-enriched immune architectures highlights potential applications in understanding prognosis, immunobiology, and therapeutic response. Further benchmarking is ongoing.
利益披露 Disclosure
S. Arunachalam, None..
O. Ospina, None..
A. C. Soupir, None..
X. Yu, None..
B. L. Fridley, None.