CITEMO ^XMBD : A flexible single-cell multimodal omics analysis framework to reveal the heterogeneity of immune cells

Abstract

Simultaneous measurement of multiple modalities in single-cell analysis, represented by CITE-seq, is a promising approach to link transcriptional changes to cellular phenotype and function, requiring new computational methods to define cellular subtypes and states based on multiple data types. Here, we design a flexible single-cell multimodal analysis framework, called CITEMO, to integrate the transcriptome and antibody-derived tags (ADT) data to capture cell heterogeneity from the multi omics perspective. CITEMO uses Principal Component Analysis (PCA) to obtain a low-dimensional representation of the transcriptome and ADT, respectively, and then employs PCA again to integrate these low-dimensional multimodal data for downstream analysis. To investigate the effectiveness of the CITEMO framework, we apply CITEMO to analyse the cell subtypes of Cord Blood Mononuclear Cells (CBMC) samples. Results show that the CITEMO framework can comprehensively analyse single-cell multimodal samples and accurately identify cell subtypes. Besides, we find some specific immune cells that co-express multiple ADT markers. To better describe the co-expression phenomenon, we introduce the co-expression entropy to measure the heterogeneous distribution of the ADT combinations. To further validate the robustness of the CITEMO framework, we analyse Human Bone Marrow Cell (HBMC) samples and identify different states of the same cell type. CITEMO has an excellent performance in identifying cell subtypes and states for multimodal omics data. We suggest that the flexible design idea of CITEMO can be an inspiration for other single-cell multimodal tasks. The complete source code and dataset of the CITEMO framework can be obtained from https://github.com/studentiz/CITEMO .

Single-cell multimodal modeling with deep parametric inference

“…To eliminate dimensional effects between metrics, data scaling is required. Transcriptome and protein data were scaled to a range of 0 to 1 by MinMaxScaler , respectively, before being fed into the DPI model [23]. x ij represents the count of the i-th feature of the j-th cell.…”

“…Transcriptome and protein data were scaled to a range of 0 to 1 by 𝑀𝑖𝑛𝑀𝑎𝑥𝑆𝑐𝑎𝑙𝑒𝑟 , respectively, before being fed into the DPI model [23].…”

Development and validation of an interpretable radiomic nomogram for severe radiation proctitis prediction in postoperative cervical cancer patients

“…Deep learning, as well as dynamical modeling, is demonstrating powerful feature extraction and modeling capabilities in various medical fields ( Li et al, 2021 ; Qian et al, 2021 ; Chen et al, 2022 ; Hu et al, 2022 ; Li Y. et al, 2022 ; Li X. et al, 2022 ). In data-driven disease research, a graph neuro network was used to predict the potential associations of disease-related metabolites ( Sun et al, 2022 ).…”

Identifying SARS-CoV-2 infected cells with scVDN

“…Single-cell RNA sequencing (scRNA-seq) has enabled the analysis of gene expression at the level of individual cells, opening up new avenues for understanding cellular heterogeneity and identifying rare cell types ( Hu et al, 2022 ; Heumos et al, 2023 ). However, the analysis of scRNA-seq data is complicated by the presence of confounding factors such as batch effects ( Zhang et al, 2022 ), cell cycle effects ( Chang et al, 2015 ), and technical noise ( Chai, 2022 ).…”