一种单细胞转录组与免疫组库的整合方法

An integrated approach for single-cell transcriptome and immune repertoire

摘要:

单细胞转录组测序（scRNA-seq）与T细胞受体免疫组库测序（scTCR-seq）是解析免疫细胞特性的两大关键技术,分别从基因表达和抗原识别维度揭示免疫系统的复杂性。然而,传统分析方法多局限于单一模态,难以有效整合两个组学所提供的互补信息。为突破这一局限,实现跨组学数据的高效融合,提出一种新型的数据整合架构scRTIA（single cell RNA and TCR integrative analysis）。该模型基于深度学习理论,以多模态变分自编码器与Transformer为核心架构,将scRNA-seq基因表达矩阵与scTCR-seq的TCR序列特征协同嵌入统一的低维潜在空间,从而构建出能同时保留转录组特征与免疫组库信息的融合细胞表征。在真实数据集上的实验验证表明,scRTIA所构建的细胞表征在细胞亚群识别方面表现出显著优越的分辨能力,能够发现传统方法难以识别的、具有特定功能状态的稀有T细胞群体。本研究通过有效深度融合转录组与免疫组库信息,突破了单模态分析的瓶颈,实现了对T细胞身份和功能的多维度刻画,在免疫相关疾病研究和精准医疗领域具有应用价值。

Abstract:

Single-cell RNA sequencing (scRNA-seq) and single-cell T-cell receptor sequencing (scTCR-seq) are pivotal for deciphering immune cell characteristics, offering complementary insights into the immune system′s complexity through gene expression and antigen recognition, respectively. However, conventional analytical methods are often confined to a single modality, hindering the effective integration of complementary information from the two omics. To overcome this limitation and achieve efficient integration of cross-omics data, this study proposes a novel data integration framework named scRTIA (single-cell RNA and TCR integrative analysis). Based on deep learning theory, the model employs a multimodal variational autoencoder and a Transformer as its core architecture, jointly embedding the scRNA-seq gene expression matrix and scTCR-seq TCR sequence features into a unified low-dimensional latent space, thereby constructing a fused cell representation that simultaneously preserves transcriptomic features and immune repertoire information. Experimental validation on real datasets demonstrates that the cell representations generated by scRTIA exhibit significantly superior resolution in identifying cell subpopulations, enabling the discovery of rare T-cell populations with specific functional states that are difficult to detect using traditional methods. By effectively integrating transcriptomic and immunome data, our work transcends the limitations of single-modal analysis and enables a multidimensional characterization of T-cell identity and function, offering valuable insights for immune-related diseases research and precision medicine.