An efficient solution to Hidden Markov Models on trees with coupled branches.

Hidden Markov Models (HMMs) are powerful tools for modeling sequential data, where the underlying states evolve in a stochastic manner and are only indirectly observable. Traditional HMM approaches are well-established for linear sequences, and have been extended to other structures such as trees. In this paper, we extend the framework of HMMs on trees to address scenarios where the tree-like structure of the data includes coupled branches -- a common feature in biological systems where entities within the same lineage exhibit dependent characteristics. We develop a dynamic programming algorithm that efficiently solves the likelihood, decoding, and parameter learning problems for tree-based HMMs with coupled branches. Our approach scales polynomially with the number of states and nodes, making it computationally feasible for a wide range of applications and does not suffer from the underflow problem. We demonstrate our algorithm by applying it to simulated data and propose self-consistency checks for validating the assumptions of the model used for inference. This work not only advances the theoretical understanding of HMMs on trees but also provides a practical tool for analyzing complex biological data where dependencies between branches cannot be ignored.

scAI-SNP: a method for inferring ancestry from single-cell data.

Collaborative efforts, such as the Human Cell Atlas, are rapidly accumulating large amounts of single-cell data. To ensure that single-cell atlases are representative of human genetic diversity, we need to determine the ancestry of the donors from whom single-cell data are generated. Self-reporting of race and ethnicity, although important, can be biased and is not always available for the datasets already collected. Here, we introduce scAI-SNP, a tool to infer ancestry directly from single-cell genomics data. To train scAI-SNP, we identified 4.5 million ancestry-informative single-nucleotide polymorphisms (SNPs) in the 1000 Genomes Project dataset across 3201 individuals from 26 population groups. For a query single-cell data set, scAI-SNP uses these ancestry-informative SNPs to compute the contribution of each of the 26 population groups to the ancestry of the donor from whom the cells were obtained. Using diverse single-cell data sets with matched whole-genome sequencing data, we show that scAI-SNP is robust to the sparsity of single-cell data, can accurately and consistently infer ancestry from samples derived from diverse types of tissues and cancer cells, and can be applied to different modalities of single-cell profiling assays, such as single-cell RNA-seq and single-cell ATAC-seq. Finally, we argue that ensuring that single-cell atlases represent diverse ancestry, ideally alongside race and ethnicity, is ultimately important for improved and equitable health outcomes by accounting for human diversity.

ProBac-seq, a bacterial single-cell RNA sequencing methodology using droplet microfluidics and large oligonucleotide probe sets.

Methods that measure the transcriptomic state of thousands of individual cells have transformed our understanding of cellular heterogeneity in eukaryotic cells since their introduction in the past decade. While simple and accessible protocols and commercial products are now available for the processing of mammalian cells, these existing technologies are incompatible with use in bacterial samples for several fundamental reasons including the absence of polyadenylation on bacterial messenger RNA, the instability of bacterial transcripts and the incompatibility of bacterial cell morphology with existing methodologies. Recently, we developed ProBac sequencing (ProBac-seq), a method that overcomes these technical difficulties and provides high-quality single-cell gene expression data from thousands of bacterial cells by using messenger RNA-specific probes. Here we provide details for designing large oligonucleotide probe sets for an organism of choice, amplifying probe sets to produce sufficient quantities for repeated experiments, adding unique molecular indexes and poly-A tails to produce finalized probes, in situ probe hybridization and single-cell encapsulation and library preparation. This protocol, from the probe amplification to the library preparation, requires ~7 d to complete. ProBac-seq offers several advantages over other methods by capturing only the desired target sequences and avoiding nondesired transcripts, such as highly abundant ribosomal RNA, thus enriching for signal that better informs on cellular state. The use of multiple probes per gene can detect meaningful single-cell signals from cells expressing transcripts to a lesser degree or those grown in minimal media and other environmentally relevant conditions in which cells are less active. ProBac-seq is also compatible with other organisms that can be profiled by in situ hybridization techniques.

Deciphering cell states and genealogies of human haematopoiesis.

The human blood system is maintained through the differentiation and massive amplification of a limited number of long-lived haematopoietic stem cells (HSCs)1. Perturbations to this process underlie diverse diseases, but the clonal contributions to human haematopoiesis and how this changes with age remain incompletely understood. Although recent insights have emerged from barcoding studies in model systems2-5, simultaneous detection of cell states and phylogenies from natural barcodes in humans remains challenging. Here we introduce an improved, single-cell lineage-tracing system based on deep detection of naturally occurring mitochondrial DNA mutations with simultaneous readout of transcriptional states and chromatin accessibility. We use this system to define the clonal architecture of HSCs and map the physiological state and output of clones. We uncover functional heterogeneity in HSC clones, which is stable over months and manifests as both differences in total HSC output and biases towards the production of different mature cell types. We also find that the diversity of HSC clones decreases markedly with age, leading to an oligoclonal structure with multiple distinct clonal expansions. Our study thus provides a clonally resolved and cell-state-aware atlas of human haematopoiesis at single-cell resolution, showing an unappreciated functional diversity of human HSC clones and, more broadly, paving the way for refined studies of clonal dynamics across a range of tissues in human health and disease.

An improved rhythmicity analysis method using Gaussian Processes detects cell-density dependent circadian oscillations in stem cells.

MOTIVATION: Detecting oscillations in time series remains a challenging problem even after decades of research. In chronobiology, rhythms (for instance in gene expression, eclosion, egg-laying, and feeding) tend to be low amplitude, display large variations amongst replicates, and often exhibit varying peak-to-peak distances (non-stationarity). Most currently available rhythm detection methods are not specifically designed to handle such datasets, and are also limited by their use of P-values in detecting oscillations. RESULTS: We introduce a new method, ODeGP (Oscillation Detection using Gaussian Processes), which combines Gaussian Process regression and Bayesian inference to incorporate measurement errors, non-uniformly sampled data, and a recently developed non-stationary kernel to improve detection of oscillations. By using Bayes factors, ODeGP models both the null (non-rhythmic) and the alternative (rhythmic) hypotheses, thus providing an advantage over P-values. Using synthetic datasets, we first demonstrate that ODeGP almost always outperforms eight commonly used methods in detecting stationary as well as non-stationary symmetric oscillations. Next, by analyzing existing qPCR datasets, we demonstrate that our method is more sensitive compared to the existing methods at detecting weak and noisy oscillations. Finally, we generate new qPCR data on mouse embryonic stem cells. Surprisingly, we discover using ODeGP that increasing cell-density results in rapid generation of oscillations in the Bmal1 gene, thus highlighting our method's ability to discover unexpected and new patterns. In its current implementation, ODeGP is meant only for analyzing single or a few time-trajectories, not genome-wide datasets. AVAILABILITY AND IMPLEMENTATION: ODeGP is available at https://github.com/Shaonlab/ODeGP.

De novo detection of somatic mutations in high-throughput single-cell profiling data sets.

Characterization of somatic mutations at single-cell resolution is essential to study cancer evolution, clonal mosaicism and cell plasticity. Here, we describe SComatic, an algorithm designed for the detection of somatic mutations in single-cell transcriptomic and ATAC-seq (assay for transposase-accessible chromatin sequence) data sets directly without requiring matched bulk or single-cell DNA sequencing data. SComatic distinguishes somatic mutations from polymorphisms, RNA-editing events and artefacts using filters and statistical tests parameterized on non-neoplastic samples. Using >2.6 million single cells from 688 single-cell RNA-seq (scRNA-seq) and single-cell ATAC-seq (scATAC-seq) data sets spanning cancer and non-neoplastic samples, we show that SComatic detects mutations in single cells accurately, even in differentiated cells from polyclonal tissues that are not amenable to mutation detection using existing methods. Validated against matched genome sequencing and scRNA-seq data, SComatic achieves F1 scores between 0.6 and 0.7 across diverse data sets, in comparison to 0.2-0.4 for the second-best performing method. In summary, SComatic permits de novo mutational signature analysis, and the study of clonal heterogeneity and mutational burdens at single-cell resolution.

Probe-based bacterial single-cell RNA sequencing predicts toxin regulation.

Clonal bacterial populations rely on transcriptional variation across individual cells to produce specialized states that increase fitness. Understanding all cell states requires studying isogenic bacterial populations at the single-cell level. Here we developed probe-based bacterial sequencing (ProBac-seq), a method that uses libraries of DNA probes and an existing commercial microfluidic platform to conduct bacterial single-cell RNA sequencing. We sequenced the transcriptome of thousands of individual bacterial cells per experiment, detecting several hundred transcripts per cell on average. Applied to Bacillus subtilis and Escherichia coli, ProBac-seq correctly identifies known cell states and uncovers previously unreported transcriptional heterogeneity. In the context of bacterial pathogenesis, application of the approach to Clostridium perfringens reveals heterogeneous expression of toxin by a subpopulation that can be controlled by acetate, a short-chain fatty acid highly prevalent in the gut. Overall, ProBac-seq can be used to uncover heterogeneity in isogenic microbial populations and identify perturbations that affect pathogenicity.

An improved rhythmicity analysis method using Gaussian Processes detects cell-density dependent circadian oscillations in stem cells.

Detecting oscillations in time series remains a challenging problem even after decades of research. In chronobiology, rhythms in time series (for instance gene expression, eclosion, egg-laying and feeding) datasets tend to be low amplitude, display large variations amongst replicates, and often exhibit varying peak-to-peak distances (non-stationarity). Most currently available rhythm detection methods are not specifically designed to handle such datasets. Here we introduce a new method, ODeGP ( O scillation De tection using G aussian P rocesses), which combines Gaussian Process (GP) regression with Bayesian inference to provide a flexible approach to the problem. Besides naturally incorporating measurement errors and non-uniformly sampled data, ODeGP uses a recently developed kernel to improve detection of non-stationary waveforms. An additional advantage is that by using Bayes factors instead of p-values, ODeGP models both the null (non-rhythmic) and the alternative (rhythmic) hypotheses. Using a variety of synthetic datasets we first demonstrate that ODeGP almost always outperforms eight commonly used methods in detecting stationary as well as non-stationary oscillations. Next, on analyzing existing qPCR datasets that exhibit low amplitude and noisy oscillations, we demonstrate that our method is more sensitive compared to the existing methods at detecting weak oscillations. Finally, we generate new qPCR time-series datasets on pluripotent mouse embryonic stem cells, which are expected to exhibit no oscillations of the core circadian clock genes. Surprisingly, we discover using ODeGP that increasing cell density can result in the rapid generation of oscillations in the Bmal1 gene, thus highlighting our method’s ability to discover unexpected patterns. In its current implementation, ODeGP (available as an R package) is meant only for analyzing single or a few time-trajectories, not genome-wide datasets.

Collective polymerase dynamics emerge from DNA supercoiling during transcription.

All biological processes ultimately come from physical interactions. The mechanical properties of DNA play a critical role in transcription. RNA polymerase can over or under twist DNA (referred to as DNA supercoiling) when it moves along a gene, resulting in mechanical stresses in DNA that impact its own motion and that of other polymerases. For example, when enough supercoiling accumulates, an isolated polymerase halts, and transcription stops. DNA supercoiling can also mediate nonlocal interactions between polymerases that shape gene expression fluctuations. Here, we construct a comprehensive model of transcription that captures how RNA polymerase motion changes the degree of DNA supercoiling, which in turn feeds back into the rate at which polymerases are recruited and move along the DNA. Surprisingly, our model predicts that a group of three or more polymerases move together at a constant velocity and sustain their motion (forming what we call a polymeton), whereas one or two polymerases would have halted. We further show that accounting for the impact of DNA supercoiling on both RNA polymerase recruitment and velocity recapitulates empirical observations of gene expression fluctuations. Finally, we propose a mechanical toggle switch whereby interactions between genes are mediated by DNA twisting as opposed to proteins. Understanding the mechanical regulation of gene expression provides new insights into how endogenous genes can interact and informs the design of new forms of engineered interactions.

A novel deep learning-based 3D cell segmentation framework for future image-based disease detection.

Cell segmentation plays a crucial role in understanding, diagnosing, and treating diseases. Despite the recent success of deep learning-based cell segmentation methods, it remains challenging to accurately segment densely packed cells in 3D cell membrane images. Existing approaches also require fine-tuning multiple manually selected hyperparameters on the new datasets. We develop a deep learning-based 3D cell segmentation pipeline, 3DCellSeg, to address these challenges. Compared to the existing methods, our approach carries the following novelties: (1) a robust two-stage pipeline, requiring only one hyperparameter; (2) a light-weight deep convolutional neural network (3DCellSegNet) to efficiently output voxel-wise masks; (3) a custom loss function (3DCellSeg Loss) to tackle the clumped cell problem; and (4) an efficient touching area-based clustering algorithm (TASCAN) to separate 3D cells from the foreground masks. Cell segmentation experiments conducted on four different cell datasets show that 3DCellSeg outperforms the baseline models on the ATAS (plant), HMS (animal), and LRP (plant) datasets with an overall accuracy of 95.6%, 76.4%, and 74.7%, respectively, while achieving an accuracy comparable to the baselines on the Ovules (plant) dataset with an overall accuracy of 82.2%. Ablation studies show that the individual improvements in accuracy is attributable to 3DCellSegNet, 3DCellSeg Loss, and TASCAN, with the 3DCellSeg demonstrating robustness across different datasets and cell shapes. Our results suggest that 3DCellSeg can serve a powerful biomedical and clinical tool, such as histo-pathological image analysis, for cancer diagnosis and grading.

Paraxial mesoderm organoids model development of human somites.

During the development of the vertebrate embryo, segmented structures called somites are periodically formed from the presomitic mesoderm (PSM) and give rise to the vertebral column. While somite formation has been studied in several animal models, it is less clear how well this process is conserved in humans. Recent progress has made it possible to study aspects of human paraxial mesoderm (PM) development such as the human segmentation clock in vitro using human pluripotent stem cells (hPSCs); however, somite formation has not been observed in these monolayer cultures. Here, we describe the generation of human PM organoids from hPSCs (termed Somitoids), which recapitulate the molecular, morphological, and functional features of PM development, including formation of somite-like structures in vitro. Using a quantitative image-based screen, we identify critical parameters such as initial cell number and signaling modulations that reproducibly yielded formation of somite-like structures in our organoid system. In addition, using single-cell RNA-sequencing and 3D imaging, we show that PM organoids both transcriptionally and morphologically resemble their in vivo counterparts and can be differentiated into somite derivatives. Our organoid system is reproducible and scalable, allowing for the systematic and quantitative analysis of human spine development and disease in vitro.

Humans are part of a group of animals called vertebrates, which are all the animals with backbones. Broadly, all vertebrates have a similar body shape with a head at one end and a left and right side that are similar to each other. Although this is not very obvious in humans, vertebrate bodies are derived from pairs of segments arranged from the head to the tail. Each of these segments or somites originates early in embryonic development. Cells from each somite then divide, grow and specialize to form bones such as the vertebrae of the vertebral column, muscles, skin, and other tissues that make up each segment. Studying different animals during embryonic development has provided insights into how somites form and grow, but it is technically difficult to do and only provides an approximate model of how somites develop in humans. Being able to make and study somites using human cells in the lab would help scientists learn more about how somite formation in humans is regulated. Budjan et al. grew human stem cells in the lab as three-dimensional structures called organoids, and used chemical signals similar to the ones produced in the embryo during development to make the cells form somites. Various combinations of signals were tested to find the best way to trigger somite formation. Once the somites formed, Budjan et al. measured them and studied their structure and the genes they used. They found that these lab-grown somites have the same size and structure as natural somites and use many of the same genes. This new organoid model provides a way to study human somite formation and development in the lab for the first time. This can provide insights into the development and evolution of humans and other animals that could then help scientists understand diseases such as the development of abnormal spinal curvature that affects around 1 in 10,000 newborns.

Transcriptional differences between JAK2-V617F and wild-type bone marrow cells in patients with myeloproliferative neoplasms.

The JAK2-V617F mutation is the most common cause of myeloproliferative neoplasms. Although experiments have revealed that this gain-of-function mutation is associated with myeloid blood cell expansion and increased production of white cells, red cells, and platelets, the transcriptional consequences of the JAK2-V617F mutation in different cellular compartments of the bone marrow have not yet been fully elucidated. To study the direct effects of JAK2-V617F on bone marrow cells in patients with myeloproliferative neoplasms, we performed joint single-cell RNA sequencing and JAK2 genotyping on CD34+-enriched cells from eight patients with newly diagnosed essential thrombocythemia or polycythemia vera. We found that the JAK2-V617F mutation increases the expression of interferon-response genes (e.g., HLAs) and the leptin receptor in hematopoietic progenitor cells. Furthermore, we sequenced a population of CD34- bone marrow monocytes and found that the JAK2 mutation increased expression of intermediate monocyte genes and the fibrocyte-associated surface protein SLAMF7 in these cells.

Integrating readout of somatic mutations in individual cells with single-cell transcriptional profiling.

In many biological applications, the readout of somatic mutations in individual cells is essential. For example, it can be used to mark individual cancer cells or identify progenies of a stem cell. Here, we present a protocol to perform single-cell RNA-seq and single-cell amplicon-seq using 10X Chromium technology. Our protocol demonstrates how to (1) isolate CD34+ progenitor cells from human bone marrow aspirate, (2) prepare single-cell amplicon libraries, and (3) analyze the libraries to assign somatic mutations to individual cells. For complete details on the use and execution of this protocol, please refer to Van Egeren et al. (2021).

Computing the Riemannian curvature of image patch and single-cell RNA sequencing data manifolds using extrinsic differential geometry.

Most high-dimensional datasets are thought to be inherently low-dimensional-that is, data points are constrained to lie on a low-dimensional manifold embedded in a high-dimensional ambient space. Here, we study the viability of two approaches from differential geometry to estimate the Riemannian curvature of these low-dimensional manifolds. The intrinsic approach relates curvature to the Laplace-Beltrami operator using the heat-trace expansion and is agnostic to how a manifold is embedded in a high-dimensional space. The extrinsic approach relates the ambient coordinates of a manifold's embedding to its curvature using the Second Fundamental Form and the Gauss-Codazzi equation. We found that the intrinsic approach fails to accurately estimate the curvature of even a two-dimensional constant-curvature manifold, whereas the extrinsic approach was able to handle more complex toy models, even when confounded by practical constraints like small sample sizes and measurement noise. To test the applicability of the extrinsic approach to real-world data, we computed the curvature of a well-studied manifold of image patches and recapitulated its topological classification as a Klein bottle. Lastly, we applied the extrinsic approach to study single-cell transcriptomic sequencing (scRNAseq) datasets of blood, gastrulation, and brain cells to quantify the Riemannian curvature of scRNAseq manifolds.

Reconstructing the Lineage Histories and Differentiation Trajectories of Individual Cancer Cells in Myeloproliferative Neoplasms.

Some cancers originate from a single mutation event in a single cell. Blood cancers known as myeloproliferative neoplasms (MPNs) are thought to originate when a driver mutation is acquired by a hematopoietic stem cell (HSC). However, when the mutation first occurs in individuals and how it affects the behavior of HSCs in their native context is not known. Here we quantified the effect of the JAK2-V617F mutation on the self-renewal and differentiation dynamics of HSCs in treatment-naive individuals with MPNs and reconstructed lineage histories of individual HSCs using somatic mutation patterns. We found that JAK2-V617F mutations occurred in a single HSC several decades before MPN diagnosis-at age 9 ± 2 years in a 34-year-old individual and at age 19 ± 3 years in a 63-year-old individual-and found that mutant HSCs have a selective advantage in both individuals. These results highlight the potential of harnessing somatic mutations to reconstruct cancer lineages.

Evaluation of Arai et al.: What Can Differences across Daughter Cells Tell Us about Stem Cell Renewal?

One snapshot of the peer review process for "Machine learning of hematopoietic stem cell divisions from paired daughter cell expression profiles reveals effects of aging on self-renewal" (Arai et al., 2020).

An Engineered CRISPR-Cas9 Mouse Line for Simultaneous Readout of Lineage Histories and Gene Expression Profiles in Single Cells.

Tracing the lineage history of cells is key to answering diverse and fundamental questions in biology. Coupling of cell ancestry information with other molecular readouts represents an important goal in the field. Here, we describe the CRISPR array repair lineage tracing (CARLIN) mouse line and corresponding analysis tools that can be used to simultaneously interrogate the lineage and transcriptomic information of single cells in vivo. This model exploits CRISPR technology to generate up to 44,000 transcribed barcodes in an inducible fashion at any point during development or adulthood, is compatible with sequential barcoding, and is fully genetically defined. We have used CARLIN to identify intrinsic biases in the activity of fetal liver hematopoietic stem cell (HSC) clones and to uncover a previously unappreciated clonal bottleneck in the response of HSCs to injury. CARLIN also allows the unbiased identification of transcriptional signatures associated with HSC activity without cell sorting.

Metabolic interactions between dynamic bacterial subpopulations.

Individual microbial species are known to occupy distinct metabolic niches within multi-species communities. However, it has remained largely unclear whether metabolic specialization can similarly occur within a clonal bacterial population. More specifically, it is not clear what functions such specialization could provide and how specialization could be coordinated dynamically. Here, we show that exponentially growing Bacillus subtilis cultures divide into distinct interacting metabolic subpopulations, including one population that produces acetate, and another population that differentially expresses metabolic genes for the production of acetoin, a pH-neutral storage molecule. These subpopulations exhibit distinct growth rates and dynamic interconversion between states. Furthermore, acetate concentration influences the relative sizes of the different subpopulations. These results show that clonal populations can use metabolic specialization to control the environment through a process of dynamic, environmentally-sensitive state-switching.

Molecular Time Sharing through Dynamic Pulsing in Single Cells.

In cells, specific regulators often compete for limited amounts of a core enzymatic resource. It is typically assumed that competition leads to partitioning of core enzyme molecules among regulators at constant levels. Alternatively, however, different regulatory species could time share, or take turns utilizing, the core resource. Using quantitative time-lapse microscopy, we analyzed sigma factor activity dynamics, and their competition for RNA polymerase, in individual Bacillus subtilis cells under energy stress. Multiple alternative sigma factors were activated in ∼1-hr pulses in stochastic and repetitive fashion. Pairwise analysis revealed that two sigma factors rarely pulse simultaneously and that some pairs are anti-correlated, indicating that RNAP utilization alternates among different sigma factors. Mathematical modeling revealed how stochastic time-sharing dynamics can emerge from pulse-generating sigma factor regulatory circuits actively competing for RNAP. Time sharing provides a mechanism for cells to dynamically control the distribution of cell states within a population. Since core molecular components are limiting in many other systems, time sharing may represent a general mode of regulation.