Dissecting heterogeneous cell populations across drug and disease conditions with PopAlign

Integrative, high-resolution analysis of single-cell gene expression across experimental conditions with PARAFAC2-RISE

Effective exploration and analysis tools are vital for the extraction of insights from single-cell data. However, current techniques for modeling single-cell studies performed across experimental conditions (e.g., samples) require restrictive assumptions or do not adequately deconvolute condition-to-condition variation from cell-to-cell variation. Here, we report that reduction and insight in single-cell exploration (RISE), an adaptation of the tensor decomposition method PARAFAC2, enables the dimensionality reduction and analysis of single-cell data across conditions. We demonstrate the benefits of RISE across distinct examples of single-cell RNA-sequencing experiments of peripheral immune cells: pharmacologic drug perturbations and systemic lupus erythematosus patient samples. RISE enables associations of gene variation patterns with patients or perturbations while connecting each coordinated change to single cells without requiring cell-type annotations. The theoretical grounding of RISE suggests a unified framework for many single-cell data modeling tasks while providing an intuitive dimensionality reduction approach for multi-sample single-cell studies across biological contexts. A record of this paper's transparent peer review process is included in the supplemental information.

scRCA: A Siamese network-based pipeline for annotating cell types using noisy single-cell RNA-seq reference data

Accurate cell type annotation is fundamentally critical for single-cell sequencing (scRNA-seq) data analysis to provide insightful knowledge of tissue-specific cell heterogeneity and cell state transition tracking. Cell type annotation is usually conducted by comparative analysis with known data (i.e., reference) - which contains a presumably accurate representation of cell types. However, this assumption is often problematic, as factors such as human errors in wet-lab experiments and methodological limitations can introduce annotation errors in the reference dataset. As current pipelines for single-cell transcriptomic analysis do not adequately consider this challenge, there is a major demand for constructing a computational pipeline that achieves high-quality cell type annotation using reference datasets containing inherent errors (referred to as "noise" in this study). Here, we built a Siamese network-based pipeline, termed scRCA, to accurately annotate cell types based on noisy reference data. To help users evaluate the reliability of scRCA annotations, an interpreter was also developed to explore the factors underlying the model's predictions. Our experiments demonstrate that, across 14 datasets, scRCA outperformed other widely adopted reference-based methods for cell type annotation. Using an independent dataset of four multiple myeloma patients, we further illustrated that scRCA can distinguish cancerous cells based on gene expression levels and identify genes closely associated with multiple myeloma through scRCA's interpretable module, providing significant information for subsequent clinical treatments. With these advancements, we anticipate that scRCA will serve as a practical reference-based approach for accurate annotating cell type annotation.

Decipher the single-cell level responses to chemotherapy in pancreatic ductal adenocarcinoma by a cross-time context graph model

Gemcitabine is commonly used for pancreatic ductal adenocarcinoma (PDAC), one of the most lethal cancer types. However, the drug resistance is a critical challenge for improving the PDAC chemotherapy. Here, we applied single-cell RNA sequencing (scRNA-seq) on PDAC patient-derived xenograft (PDX) models to study the complex cellular responses related to the gemcitabine resistances. To reconstruct dynamic tumor cell responses from these static scRNA-seq snapshots, we proposed scConGraph, a scalable bi-layer graph model that can efficiently integrate cross-time context information. Based on scConGraph, we observed that stemness and endoplasmic reticulum stress contribute to intrinsic resistance. As for acquired resistance, cancer cells may resist or evade gemcitabine treatment by activating the cell cycle, entering quiescence, or inducing epithelial-mesenchymal transition. Notably, GDF15 exhibited recurrent and significant upregulations among acquired-resistance cell subpopulations. Experimental validation confirmed that inhibiting GDF15 sensitizes tumor cells to gemcitabine, suggesting a potential target for gemcitabine-induced chemoresistance.

Cell-cycle dependence of bursty gene expression: insights from fitting mechanistic models to single-cell RNA-seq data

Bursty gene expression is characterized by two intuitive parameters, burst frequency and burst size, the cell-cycle dependence of which has not been extensively profiled at the transcriptome level. In this study, we estimate the burst parameters per allele in the G1 and G2/M cell-cycle phases for thousands of mouse genes by fitting mechanistic models of gene expression to messenger RNA count data, obtained by sequencing of single cells whose cell-cycle position has been inferred using a deep-learning method. We find that upon DNA replication, the median burst frequency approximately halves, while the burst size remains mostly unchanged. Genome-wide distributions of the burst parameter ratios between the G2/M and G1 phases are broad, indicating substantial heterogeneity in transcriptional regulation. We also observe a significant negative correlation between the burst frequency and size ratios, suggesting that regulatory processes do not independently control the burst parameters. We show that to accurately estimate the burst parameter ratios, mechanistic models must explicitly account for gene copy number variation and extrinsic noise due to the coupling of transcription to cell age across the cell cycle, but corrections for technical noise due to imperfect capture of RNA molecules in sequencing experiments are less critical.

Integrative, high-resolution analysis of single cell gene expression across experimental conditions with PARAFAC2-RISE

Effective and scalable exploration and analysis tools are vital for the extraction of insights from large-scale single-cell data. However, current techniques for modeling single-cell studies performed across experimental conditions (e.g., samples, perturbations, or patients) require restrictive assumptions, lack flexibility, or do not adequately deconvolute condition-to-condition variation from cell-to-cell variation. Here, we report that Reduction and Insight in Single-cell Exploration (RISE), an adaptation of the tensor decomposition method PARAFAC2, enables the dimensionality reduction and analysis of single-cell data across conditions. We demonstrate the benefits of RISE across two distinct examples of single-cell RNA-sequencing experiments of peripheral immune cells: pharmacologic drug perturbations and systemic lupus erythematosus (SLE) patient samples. RISE enables straightforward associations of gene variation patterns with specific patients or perturbations, while connecting each coordinated change to single cells without requiring cell type annotations. The theoretical grounding of RISE suggests a unified framework for many single-cell data modeling tasks. Thus, RISE provides an intuitive universal dimensionality reduction approach for multi-sample single-cell studies across diverse biological contexts.

scStateDynamics: deciphering the drug-responsive tumor cell state dynamics by modeling single-cell level expression changes

Understanding tumor cell heterogeneity and plasticity is crucial for overcoming drug resistance. Single-cell technologies enable analyzing cell states at a given condition, but catenating static cell snapshots to characterize dynamic drug responses remains challenging. Here, we propose scStateDynamics, an algorithm to infer tumor cell state dynamics and identify common drug effects by modeling single-cell level gene expression changes. Its reliability is validated on both simulated and lineage tracing data. Application to real tumor drug treatment datasets identifies more subtle cell subclusters with different drug responses beyond static transcriptome similarity and disentangles drug action mechanisms from the cell-level expression changes.

Biophysically interpretable inference of cell types from multimodal sequencing data

Multimodal, single-cell genomics technologies enable simultaneous measurement of multiple facets of DNA and RNA processing in the cell. This creates opportunities for transcriptome-wide, mechanistic studies of cellular processing in heterogeneous cell populations, such as regulation of cell fate by transcriptional stochasticity or tumor proliferation through aberrant splicing dynamics. However, current methods for determining cell types or 'clusters' in multimodal data often rely on ad hoc approaches to balance or integrate measurements, and assumptions ignoring inherent properties of the data. To enable interpretable and consistent cell cluster determination, we present meK-means (mechanistic K-means) which integrates modalities through a unifying model of transcription to learn underlying, shared biophysical states. With meK-means we can cluster cells with nascent and mature mRNA measurements, utilizing the causal, physical relationships between these modalities. This identifies shared transcription dynamics across cells, which induce the observed molecule counts, and provides an alternative definition for 'clusters' through the governing parameters of cellular processes.

Fatecode enables cell fate regulator prediction using classification-supervised autoencoder perturbation

Cell reprogramming, which guides the conversion between cell states, is a promising technology for tissue repair and regeneration, with the ultimate goal of accelerating recovery from diseases or injuries. To accomplish this, regulators must be identified and manipulated to control cell fate. We propose Fatecode, a computational method that predicts cell fate regulators based only on single-cell RNA sequencing (scRNA-seq) data. Fatecode learns a latent representation of the scRNA-seq data using a deep learning-based classification-supervised autoencoder and then performs in silico perturbation experiments on the latent representation to predict genes that, when perturbed, would alter the original cell type distribution to increase or decrease the population size of a cell type of interest. We assessed Fatecode's performance using simulations from a mechanistic gene-regulatory network model and scRNA-seq data mapping blood and brain development of different organisms. Our results suggest that Fatecode can detect known cell fate regulators from single-cell transcriptomics datasets.

Stochastic Modeling of Biophysical Responses to Perturbation

Recent advances in high-throughput, multi-condition experiments allow for genome-wide investigation of how perturbations affect transcription and translation in the cell across multiple biological entities or modalities, from chromatin and mRNA information to protein production and spatial morphology. This presents an unprecedented opportunity to unravel how the processes of DNA and RNA regulation direct cell fate determination and disease response. Most methods designed for analyzing large-scale perturbation data focus on the observational outcomes, e.g., expression; however, many potential transcriptional mechanisms, such as transcriptional bursting or splicing dynamics, can underlie these complex and noisy observations. In this analysis, we demonstrate how a stochastic biophysical modeling approach to interpreting high-throughout perturbation data enables deeper investigation of the 'how' behind such molecular measurements. Our approach takes advantage of modalities already present in data produced with current technologies, such as nascent and mature mRNA measurements, to illuminate transcriptional dynamics induced by perturbation, predict kinetic behaviors in new perturbation settings, and uncover novel populations of cells with distinct kinetic responses to perturbation.

D-SPIN constructs gene regulatory network models from multiplexed scRNA-seq data revealing organizing principles of cellular perturbation response

Gene regulatory networks within cells modulate the expression of the genome in response to signals and changing environmental conditions. Reconstructions of gene regulatory networks can reveal the information processing and control principles used by cells to maintain homeostasis and execute cell-state transitions. Here, we introduce a computational framework, D-SPIN, that generates quantitative models of gene regulatory networks from single-cell mRNA-seq datasets collected across thousands of distinct perturbation conditions. D-SPIN models the cell as a collection of interacting gene-expression programs, and constructs a probabilistic model to infer regulatory interactions between gene-expression programs and external perturbations. Using large Perturb-seq and drug-response datasets, we demonstrate that D-SPIN models reveal the organization of cellular pathways, sub-functions of macromolecular complexes, and the logic of cellular regulation of transcription, translation, metabolism, and protein degradation in response to gene knockdown perturbations. D-SPIN can also be applied to dissect drug response mechanisms in heterogeneous cell populations, elucidating how combinations of immunomodulatory drugs can induce novel cell states through additive recruitment of gene expression programs. D-SPIN provides a computational framework for constructing interpretable models of gene-regulatory networks to reveal principles of cellular information processing and physiological control.

Learning single-cell perturbation responses using neural optimal transport

Understanding and predicting molecular responses in single cells upon chemical, genetic or mechanical perturbations is a core question in biology. Obtaining single-cell measurements typically requires the cells to be destroyed. This makes learning heterogeneous perturbation responses challenging as we only observe unpaired distributions of perturbed or non-perturbed cells. Here we leverage the theory of optimal transport and the recent advent of input convex neural architectures to present CellOT, a framework for learning the response of individual cells to a given perturbation by mapping these unpaired distributions. CellOT outperforms current methods at predicting single-cell drug responses, as profiled by scRNA-seq and a multiplexed protein-imaging technology. Further, we illustrate that CellOT generalizes well on unseen settings by (1) predicting the scRNA-seq responses of holdout patients with lupus exposed to interferon-β and patients with glioblastoma to panobinostat; (2) inferring lipopolysaccharide responses across different species; and (3) modeling the hematopoietic developmental trajectories of different subpopulations.

Recovery of missing single-cell RNA-sequencing data with optimized transcriptomic references

Single-cell RNA-sequencing (scRNA-seq) is an indispensable tool for characterizing cellular diversity and generating hypotheses throughout biology. Droplet-based scRNA-seq datasets often lack expression data for genes that can be detected with other methods. Here we show that the observed sensitivity deficits stem from three sources: (1) poor annotation of 3' gene ends; (2) issues with intronic read incorporation; and (3) gene overlap-derived read loss. We show that missing gene expression data can be recovered by optimizing the reference transcriptome for scRNA-seq through recovering false intergenic reads, implementing a hybrid pre-mRNA mapping strategy and resolving gene overlaps. We demonstrate, with a diverse collection of mouse and human tissue data, that reference optimization can substantially improve cellular profiling resolution and reveal missing cell types and marker genes. Our findings argue that transcriptomic references need to be optimized for scRNA-seq analysis and warrant a reanalysis of previously published datasets and cell atlases.

Biophysically Interpretable Inference of Cell Types from Multimodal Sequencing Data

Multimodal, single-cell genomics technologies enable simultaneous capture of multiple facets of DNA and RNA processing in the cell. This creates opportunities for transcriptome-wide, mechanistic studies of cellular processing in heterogeneous cell types, with applications ranging from inferring kinetic differences between cells, to the role of stochasticity in driving heterogeneity. However, current methods for determining cell types or 'clusters' present in multimodal data often rely on ad hoc or independent treatment of modalities, and assumptions ignoring inherent properties of the count data. To enable interpretable and consistent cell cluster determination from multimodal data, we present meK-Means (mechanistic K-Means) which integrates modalities and learns underlying, shared biophysical states through a unifying model of transcription. In particular, we demonstrate how meK-Means can be used to cluster cells from unspliced and spliced mRNA count modalities. By utilizing the causal, physical relationships underlying these modalities, we identify shared transcriptional kinetics across cells, which induce the observed gene expression profiles, and provide an alternative definition for 'clusters' through the governing parameters of cellular processes.

The specious art of single-cell genomics

Dimensionality reduction is standard practice for filtering noise and identifying relevant features in large-scale data analyses. In biology, single-cell genomics studies typically begin with reduction to 2 or 3 dimensions to produce "all-in-one" visuals of the data that are amenable to the human eye, and these are subsequently used for qualitative and quantitative exploratory analysis. However, there is little theoretical support for this practice, and we show that extreme dimension reduction, from hundreds or thousands of dimensions to 2, inevitably induces significant distortion of high-dimensional datasets. We therefore examine the practical implications of low-dimensional embedding of single-cell data and find that extensive distortions and inconsistent practices make such embeddings counter-productive for exploratory, biological analyses. In lieu of this, we discuss alternative approaches for conducting targeted embedding and feature exploration to enable hypothesis-driven biological discovery.

SimCH: simulation of single-cell RNA sequencing data by modeling cellular heterogeneity at gene expression level

Single-cell ribonucleic acid (RNA) sequencing (scRNA-seq) has been a powerful technology for transcriptome analysis. However, the systematic validation of diverse computational tools used in scRNA-seq analysis remains challenging. Here, we propose a novel simulation tool, termed as Simulation of Cellular Heterogeneity (SimCH), for the flexible and comprehensive assessment of scRNA-seq computational methods. The Gaussian Copula framework is recruited to retain gene coexpression of experimental data shown to be associated with cellular heterogeneity. The synthetic count matrices generated by suitable SimCH modes closely match experimental data originating from either homogeneous or heterogeneous cell populations and either unique molecular identifier (UMI)-based or non-UMI-based techniques. We demonstrate how SimCH can benchmark several types of computational methods, including cell clustering, discovery of differentially expressed genes, trajectory inference, batch correction and imputation. Moreover, we show how SimCH can be used to conduct power evaluation of cell clustering methods. Given these merits, we believe that SimCH can accelerate single-cell research.

Current perspectives on interethnic variability in multiple myeloma: Single cell technology, population pharmacogenetics and molecular signal transduction

•

This review discusses the emerging single cell technologies and applications in Multiple myeloma (MM), population pharmacogenetics of MM, resistance to chemotherapy, genetic determinants of drug-induced toxicity, molecular signal transduction.

•

The role(s) of epigenetics and noncoding RNAs including microRNAs (miRNAs) and long noncoding RNAs (lncRNAs) that influence the risk and severity of MM are also discussed.

•

It is understood that ethnic component acts as a driver of variable response to chemotherapy in different sub-populations globally.

•

This review augments our understanding of genetic variability in ‘myelomagenesis’ and drug-induced toxicity, myeloma microenvironment at the molecular and cellular level, and developing precision medicine strategies to combat this malignancy.

•

The emerging single cell technologies hold great promise for enhancing our understanding of MM tumor heterogeneity and clonal diversity.

Multiple myeloma (MM) is an aggressive cancer characterised by malignancy of the plasma cells and a rising global incidence. The gold standard for optimum response is aggressive chemotherapy followed by autologous stem cell transplantation (ASCT). However, majority of the patients are above 60 years and this presents the clinician with complications such as ineligibility for ASCT, frailty, drug-induced toxicity and differential/partial response to treatment. The latter is partly driven by heterogenous genotypes of the disease in different subpopulations. In this review, we discuss emerging single cell technologies and applications in MM, population pharmacogenetics of MM, resistance to chemotherapy, genetic determinants of drug-induced toxicity, molecular signal transduction, as well as the role(s) played by epigenetics and noncoding RNAs including microRNAs (miRNAs) and long noncoding RNAs (lncRNAs) that influence the risk and severity of the disease. Taken together, our discussions further our understanding of genetic variability in ‘myelomagenesis’ and drug-induced toxicity, augment our understanding of the myeloma microenvironment at the molecular and cellular level and provide a basis for developing precision medicine strategies to combat this malignancy.

Minimal gene set discovery in single-cell mRNA-seq datasets with ActiveSVM

Sequencing costs currently prohibit the application of single-cell mRNA-seq to many biological and clinical analyses. Targeted single-cell mRNA-sequencing reduces sequencing costs by profiling reduced gene sets that capture biological information with a minimal number of genes. Here we introduce an active learning method that identifies minimal but highly informative gene sets that enable the identification of cell types, physiological states and genetic perturbations in single-cell data using a small number of genes. Our active feature selection procedure generates minimal gene sets from single-cell data by employing an active support vector machine (ActiveSVM) classifier. We demonstrate that ActiveSVM feature selection identifies gene sets that enable ~90% cell-type classification accuracy across, for example, cell atlas and disease-characterization datasets. The discovery of small but highly informative gene sets should enable reductions in the number of measurements necessary for application of single-cell mRNA-seq to clinical tests, therapeutic discovery and genetic screens.

Whole-animal multiplexed single-cell RNA-seq reveals transcriptional shifts across Clytia medusa cell types

We present an organism-wide, transcriptomic cell atlas of the hydrozoan medusa Clytia hemisphaerica and describe how its component cell types respond to perturbation. Using multiplexed single-cell RNA sequencing, in which individual animals were indexed and pooled from control and perturbation conditions into a single sequencing run, we avoid artifacts from batch effects and are able to discern shifts in cell state in response to organismal perturbations. This work serves as a foundation for future studies of development, function, and regeneration in a genetically tractable jellyfish species. Moreover, we introduce a powerful workflow for high-resolution, whole-animal, multiplexed single-cell genomics that is readily adaptable to other traditional or nontraditional model organisms.

Reconstructing aspects of human embryogenesis with pluripotent stem cells

Understanding human development is of fundamental biological and clinical importance. Despite its significance, mechanisms behind human embryogenesis remain largely unknown. Here, we attempt to model human early embryo development with expanded pluripotent stem cells (EPSCs) in 3-dimensions. We define a protocol that allows us to generate self-organizing cystic structures from human EPSCs that display some hallmarks of human early embryogenesis. These structures mimic polarization and cavitation characteristic of pre-implantation development leading to blastocyst morphology formation and the transition to post-implantation-like organization upon extended culture. Single-cell RNA sequencing of these structures reveals subsets of cells bearing some resemblance to epiblast, hypoblast and trophectoderm lineages. Nevertheless, significant divergences from natural blastocysts persist in some key markers, and signalling pathways point towards ways in which morphology and transcriptional-level cell identities may diverge in stem cell models of the embryo. Thus, this stem cell platform provides insights into the design of stem cell models of embryogenesis.

Machine learning for perturbational single-cell omics

Cell biology is fundamentally limited in its ability to collect complete data on cellular phenotypes and the wide range of responses to perturbation. Areas such as computer vision and speech recognition have addressed this problem of characterizing unseen or unlabeled conditions with the combined advances of big data, deep learning, and computing resources in the past 5 years. Similarly, recent advances in machine learning approaches enabled by single-cell data start to address prediction tasks in perturbation response modeling. We first define objectives in learning perturbation response in single-cell omics; survey existing approaches, resources, and datasets (https://github.com/theislab/sc-pert); and discuss how a perturbation atlas can enable deep learning models to construct an informative perturbation latent space. We then examine future avenues toward more powerful and explainable modeling using deep neural networks, which enable the integration of disparate information sources and an understanding of heterogeneous, complex, and unseen systems.