Spectral clustering of single-cell multi-omics data on multilayer graphs.

Multilayer graphs are commonly used for representing different relations between entities and handling heterogeneous data processing tasks. Nonstandard multilayer graph clustering methods are needed for assigning clusters to a common multilayer node set and for combining information from each layer. This paper presents a multilayer spectral graph clustering (SGC) framework that performs convex layer aggregation. Under a multilayer signal-plus-noise model, we provide a phase transition analysis of clustering reliability. Moreover, we use the phase transition criterion to propose a multilayer iterative model order selection algorithm (MIMOSA) for multilayer SGC, which features automated cluster assignment and layer weight adaptation, and provides statistical clustering reliability guarantees. Numerical simulations on synthetic multilayer graphs verify the phase transition analysis, and experiments on real-world multilayer graphs show that MIMOSA is competitive or better than other clustering methods.

Whole exome and genome sequencing, coupled with refined bioinformatic pipelines, have enabled improved diagnostic yields for individuals with Mendelian conditions and have led to the rapid identification of novel syndromes. For many Mendelian neurodevelopmental disorders (NDDs), there is a lack of pre-existing model systems for mechanistic work. Thus, it is critical for translational researchers to have an accessible phenotype- and genotype-informed approach for model system selection. Single-cell RNA sequencing data can be informative in such an approach, as it can indicate which cell types express a gene of interest at the highest levels across time. For Mendelian NDDs, such data for the developing human brain is especially useful. A valuable single-cell RNA sequencing dataset of the second trimester developing human brain was produced by Bhaduri et al in 2021, but access to these data can be limited by computing power and the learning curve of single-cell data analysis. To reduce these barriers for translational research on Mendelian NDDs, we have built the web-based tool, Neurodevelopment in Trimester 2 - VIsualization of Single cell Data Online Tool (NeuroTri2-VISDOT), for exploring this single-cell dataset, and we have employed it in several different settings to demonstrate its utility for the translational research community.

With the development of single-cell sequencing technology, it is a hot research topic to identify the cell types using single-cell sequencing data, and many single-cell clustering algorithms have been developed to study this issue. These methods capture partial information of single-cell sequencing data, and obtain the different performance on the same data set. Combining these different results into one can improve the accuracy and validity. Here, we proposed ECBN, Ensemble Clustering based on Bayesian Network. ECBN can ensemble several different results of state-of-the-art single cell clustering methods, such as Seurat, CIDR, SC3 and t-SNE+k-means, and generate a more optimal clustering result through Bayesian network. Experiments are carried on the 5 single cell data sets and compared with 4 individual single cell clustering methods and 3 integrative methods. The size of experiment data sets ranges from 822 to 3605 and the results show that our method can achieve good performance. Moreover, ECBN can also use the graphical regularization to lighten the limitation which is generated by the different basis results.

Drug response prediction at the single-cell level provides guidance for drug development and treatment. Existing methods typically rely on integrating data from bulk RNA sequencing and single-cell RNA sequencing, transferring overall cell line drug response labels to individual cells. However, the inherent differences between these data types and the assumption that the cell line response represents each cell's response can affect prediction accuracy. This study proposes a heterogeneous network transfer learning model, scXDR, for drug response prediction across single-cell datasets. The model uses heterogeneous networks to integrate features and associations among drugs, genes, and cells, performing message passing, feature and structure alignment, structure reconstruction, and drug-cell score prediction. Experimental results show that scXDR outperforms methods transferring from bulk data to single-cell data and methods transferring from single-cell data to single-cell data in multiple scenarios, achieving excellent performance at individual cell and cell group levels. The importance of the model components is also confirmed. Case studies include outcome prediction for cells under drug holiday treatment, melanoma drug screening, multi-tumor drug response analysis, and predictions for approved drugs and potential drug combinations for patient cells. The research strategy and results of scXDR provide references for precision treatment at the cell level.

The development of single-cell transcriptomic technologies yields large datasets comprising multimodal informations, such as transcriptomes and immunophenotypes. Despite the current explosion of methods for pre-processing and integrating multimodal single-cell data, there is currently no user-friendly software to display easily and simultaneously both immunophenotype and transcriptome-based UMAP/t-SNE plots from the pre-processed data. Here, we introduce Single-Cell Virtual Cytometer, an open-source software for flow cytometry-like visualization and exploration of pre-processed multi-omics single cell datasets. Using an original CITE-seq dataset of PBMC from an healthy donor, we illustrate its use for the integrated analysis of transcriptomes and epitopes of functional maturation in human peripheral T lymphocytes. So this free and open-source algorithm constitutes a unique resource for biologists seeking for a user-friendly analytic tool for multimodal single cell datasets.

The well-known method of spectral clustering for 3D shape datasets is revisited. The graph construction by NNLS (Nonnegative Least Squares) coding technique is at the core of our method. Using graph-based encoding techniques and well-known shape descriptors, a framework is presented here which is applied to 3D shape clustering. Provided a shape database, a graph is first constructed. Weights in the graph are calculated so as to approximate each shape as a sparse linear combination of the remaining dataset objects. This framework is further enhanced by using the multilayer graphs process combining NNLS with L2graph. The L2graph is a sparse similarity graph, which conveys complementary information to NNLS coding. A criterion for the complementary action of the two graphs in terms of graph distance is also presented. Experimental results conducted on SHREC10, SHREC11 and SCHREC15 datasets validate the excellent performance of our clustering framework.

Large single-cell RNA datasets have contributed to unprecedented biological insight. Often, these take the form of cell atlases and serve as a reference for automating cell labeling of newly sequenced samples. Yet, classification algorithms have lacked the capacity to accurately annotate cells, particularly in complex datasets. Here we present SIMS (Scalable, Interpretable Machine Learning for Single-Cell), an end-to-end data-efficient machine learning pipeline for discrete classification of single-cell data that can be applied to new datasets with minimal coding. We benchmarked SIMS against common single-cell label transfer tools and demonstrated that it performs as well or better than state of the art algorithms. We then use SIMS to classify cells in one of the most complex tissues: the brain. We show that SIMS classifies cells of the adult cerebral cortex and hippocampus at a remarkably high accuracy. This accuracy is maintained in trans-sample label transfers of the adult human cerebral cortex. We then apply SIMS to classify cells in the developing brain and demonstrate a high level of accuracy at predicting neuronal subtypes, even in periods of fate refinement, shedding light on genetic changes affecting specific cell types across development. Finally, we apply SIMS to single cell datasets of cortical organoids to predict cell identities and unveil genetic variations between cell lines. SIMS identifies cell-line differences and misannotated cell lineages in human cortical organoids derived from different pluripotent stem cell lines. When cell types are obscured by stress signals, label transfer from primary tissue improves the accuracy of cortical organoid annotations, serving as a reliable ground truth. Altogether, we show that SIMS is a versatile and robust tool for cell-type classification from single-cell datasets.

Lymphangioleiomyomatosis (LAM) is a rare lung disease of women, causing cystic remodelling of the lung and progressive respiratory failure. The cellular composition, microenvironment and cellular interactions within the LAM lesion...

Multimodal single-cell omics technologies enable multiple molecular programs to be simultaneously profiled at a global scale in individual cells, creating opportunities to study biological systems at a resolution that was previously inaccessible. However, the analysis of multimodal single-cell omics data is challenging due to the lack of methods that can integrate across multiple data modalities generated from such technologies. Here, we present Matilda, a multi-task learning method for integrative analysis of multimodal single-cell omics data. By leveraging the interrelationship among tasks, Matilda learns to perform data simulation, dimension reduction, cell type classification, and feature selection in a single unified framework. We compare Matilda with other state-of-the-art methods on datasets generated from some of the most popular multimodal single-cell omics technologies. Our results demonstrate the utility of Matilda for addressing multiple key tasks on integrative multimodal single-cell omics data analysis. Matilda is implemented in Pytorch and is freely available from https://github.com/PYangLab/Matilda.

Anti-CD19 CAR-T cell immunotherapy is a hopeful treatment option for patients with B cell lymphomas, however it copes with partly severe adverse effects like neurotoxicity. Single-cell resolved molecular data sets in combination with clinical parametrization allow for comprehensive characterization of cellular subpopulations, their transcriptomic states, and their relation to the adverse effects. We here present a re-analysis of single-cell RNA sequencing data of 24 patients comprising more than 130,000 cells with focus on cellular states and their association to immune cell related neurotoxicity. For this, we developed a single-cell data portraying workflow to disentangle the transcriptional state space with single-cell resolution and its analysis in terms of modularly-composed cellular programs. We demonstrated capabilities of single-cell data portraying to disentangle transcriptional states using intuitive visualization, functional mining, molecular cell stratification, and variability analyses. Our analysis revealed that the T cell composition of the patient’s infusion product as well as the spectrum of their transcriptional states of cells derived from patients with low ICANS grade do not markedly differ from those of cells from high ICANS patients, while the relative abundancies, particularly that of cycling cells, of LAG3-mediated exhaustion and of CAR positive cells, vary. Our study provides molecular details of the transcriptomic landscape with possible impact to overcome neurotoxicity.

Enhancers play an important role in transcriptional regulation by modulating gene expression from distal genomic locations. Although single-cell ATAC and RNA sequencing (scATAC/RNA-seq) data have been leveraged to infer enhancer-gene links, establishing regulatory links between enhancers and their target genes remains a challenge due to the absence of chromatin conformation information. Here, we present SCEG-HiC, a machine learning method based on weighted graphical lasso, which decodes enhancer-gene links from single-cell multi-omics data by integrating bulk average Hi-C as prior knowledge. SCEG-HiC supports both paired scATAC/RNA-seq and scATAC-only inputs, improving prediction accuracy while retaining context-specific correlations and enabling the discovery of biologically relevant links. Comprehensive evaluation across 10 human and mouse single-cell multi-omics datasets shows that SCEG-HiC outperforms existing single-cell models. Application of SCEG-HiC to COVID-19 datasets illustrates its capacity to more reliably reconstruct gene regulatory networks underlying disease severity, and elucidate functional associations between noncoding variants and their putative target genes. SCEG-HiC is freely available as an open-source and user-friendly R package, facilitating broad applications in regulatory genomics research.

The prevalence of single-cell multi-omics datasets calls for automated cell type annotation methods that can characterize novel cell states. We developed Phi-Space, a computational framework for the continuous phenotyping of single-cell multi-omics data. We adopt a highly versatile modeling strategy to characterize query cell identity in a low-dimensional phenotype space, defined by reference phenotypes. The phenotype space embedding enables various downstream analyses, including insightful visualizations, clustering, and cell type labeling. Phi-Space is applicable to a wide range analytical tasks beyond cell type transfer. Its ability to model complex phenotypic variations will facilitate biological discoveries from different omics types.Supplementary informationThe online version contains supplementary material available at 10.1186/s13059-025-03755-8.

Recent advancements in single-cell multi-omics technologies have significantly improved our ability to explore cellular heterogeneity at an unprecedented resolution. These innovations enable the simultaneous profiling of genomic, transcriptomic, proteomic, and epigenetic data at the single-cell level, providing comprehensive insights into cellular states and their regulatory mechanisms. However, integrating multi-omics data remains challenging due to its high dimensionality, technical noise, and biological complexity. To address these challenges, we introduce scWDAC (single-cell weighted distance adaptive clustering), a novel clustering method for single-cell multi-omics data. scWDAC utilizes a weighted distance penalty and adaptive graph regularization to effectively integrate multiple omics layers. Key innovations of scWDAC include using a weighted distance penalty to capture cell-to-cell similarities across different omics modalities, and applying adaptive graph regularization on a consensus matrix to enforce cross-modal consistency. The framework optimizes both global consistency and local accuracy, ensuring a robust exploration of cellular structures across all omics layers. The effectiveness of scWDAC is evaluated through extensive experiments on ten paired single-cell multi-omics datasets. The results demonstrate that scWDAC outperforms existing clustering methods in terms of clustering accuracy, robustness to noise, and biological interpretability.

Spectral algorithms are graph partitioning algorithms that partition a node set of a graph into groups by using a spectral embedding map. Clustering techniques based on the algorithms are referred to as spectral clustering and are widely used in data analysis. To gain a better understanding of why spectral clustering is successful, Peng et al. (2015) and Kolev and Mehlhorn (2016) studied the behavior of a certain type of spectral algorithm for a class of graphs, called well-clustered graphs. Specifically, they put an assumption on graphs and showed the performance guarantee of the spectral algorithm under it. The algorithm they studied used the spectral embedding map developed by Shi and Malic (2000). In this paper, we improve on their results, giving a better performance guarantee under a weaker assumption. We also evaluate the performance of the spectral algorithm with the spectral embedding map developed by Ng et al. (2001).

Constrained spectral embedding for K-way data clustering