Fast searches of large collections of single-cell data using scfind

Mining alternative splicing patterns in scRNA-seq data using scASfind

Single-cell RNA-seq (scRNA-seq) is widely used for transcriptome profiling, but most analyses focus on gene-level events, with less attention devoted to alternative splicing. Here, we present scASfind, a novel computational method to allow for quantitative analysis of cell type-specific splicing events using full-length scRNA-seq data. ScASfind utilizes an efficient data structure to store the percent spliced-in value for each splicing event. This makes it possible to exhaustively search for patterns among all differential splicing events, allowing us to identify marker events, mutually exclusive events, and events involving large blocks of exons that are specific to one or more cell types.

Single-cell transcriptomics identifies perturbed molecular pathways in midbrain organoids using α-synuclein triplication Parkinson's disease patient-derived iPSCs

Three-dimensional (3D) brain organoids provide a platform to study brain development, cellular coordination, and disease using human tissue. Here, we generate midbrain dopaminergic (mDA) organoids from induced pluripotent stem cells (iPSC) from healthy and Parkinson's Disease (PD) donors and assess them as a human PD model using single-cell RNAseq. We characterize cell types in our organoid cultures and analyze our model's Dopamine (DA) neurons using cytotoxic and genetic stressors. Our study provides the first in-depth, single-cell analysis of SNCA triplication and shows evidence for molecular dysfunction in oxidative phosphorylation, translation, and ER protein-folding in DA neurons. We perform an in-silico identification of rotenone-sensitive DA neurons and characterization of corresponding transcriptomic profiles associated with synaptic signalling and cholesterol biosynthesis. Finally, we show a novel chimera organoid model from healthy and PD iPSCs allowing the study of DA neurons from different individuals within the same tissue.

Matching queried single-cell open-chromatin profiles to large pools of single-cell transcriptomes and epigenomes for reference supported analysis

The true benefits of large single-cell transcriptome and epigenome data sets can be realized only with the development of new approaches and search tools for annotating individual cells. Matching a single-cell epigenome profile to a large pool of reference cells remains a major challenge. Here, we present scEpiSearch, which enables searching, comparison, and independent classification of single-cell open-chromatin profiles against a large reference of single-cell expression and open-chromatin data sets. Across performance benchmarks, scEpiSearch outperformed multiple methods in accuracy of search and low-dimensional coembedding of single-cell profiles, irrespective of platforms and species. Here we also demonstrate the unconventional utilities of scEpiSearch by applying it on single-cell epigenome profiles of K562 cells and samples from patients with acute leukaemia to reveal different aspects of their heterogeneity, multipotent behavior, and dedifferentiated states. Applying scEpiSearch on our single-cell open-chromatin profiles from embryonic stem cells (ESCs), we identified ESC subpopulations with more activity and poising for endoplasmic reticulum stress and unfolded protein response. Thus, scEpiSearch solves the nontrivial problem of amalgamating information from a large pool of single cells to identify and study the regulatory states of cells using their single-cell epigenomes.

Nonnegative spatial factorization applied to spatial genomics

Nonnegative matrix factorization (NMF) is widely used to analyze high-dimensional count data because, in contrast to real-valued alternatives such as factor analysis, it produces an interpretable parts-based representation. However, in applications such as spatial transcriptomics, NMF fails to incorporate known structure between observations. Here, we present nonnegative spatial factorization (NSF), a spatially-aware probabilistic dimension reduction model based on transformed Gaussian processes that naturally encourages sparsity and scales to tens of thousands of observations. NSF recovers ground truth factors more accurately than real-valued alternatives such as MEFISTO in simulations, and has lower out-of-sample prediction error than probabilistic NMF on three spatial transcriptomics datasets from mouse brain and liver. Since not all patterns of gene expression have spatial correlations, we also propose a hybrid extension of NSF that combines spatial and nonspatial components, enabling quantification of spatial importance for both observations and features. A TensorFlow implementation of NSF is available from https://github.com/willtownes/nsf-paper .

scHumanNet: a single-cell network analysis platform for the study of cell-type specificity of disease genes

A major challenge in single-cell biology is identifying cell-type-specific gene functions, which may substantially improve precision medicine. Differential expression analysis of genes is a popular, yet insufficient approach, and complementary methods that associate function with cell type are required. Here, we describe scHumanNet (https://github.com/netbiolab/scHumanNet), a single-cell network analysis platform for resolving cellular heterogeneity across gene functions in humans. Based on cell-type-specific gene networks (CGNs) constructed under the guidance of the HumanNet reference interactome, scHumanNet displayed higher functional relevance to the cellular context than CGNs built by other methods on single-cell transcriptome data. Cellular deconvolution of gene signatures based on network compactness across cell types revealed breast cancer prognostic markers associated with T cells. scHumanNet could also prioritize genes associated with particular cell types using CGN centrality and identified the differential hubness of CGNs between disease and healthy conditions. We demonstrated the usefulness of scHumanNet by uncovering T-cell-specific functional effects of GITR, a prognostic gene for breast cancer, and functional defects in autism spectrum disorder genes specific for inhibitory neurons. These results suggest that scHumanNet will advance our understanding of cell-type specificity across human disease genes.

Production of human entorhinal stellate cell-like cells by forward programming shows an important role of Foxp1 in reprogramming

Stellate cells are principal neurons in the entorhinal cortex that contribute to spatial processing. They also play a role in the context of Alzheimer's disease as they accumulate Amyloid beta early in the disease. Producing human stellate cells from pluripotent stem cells would allow researchers to study early mechanisms of Alzheimer's disease, however, no protocols currently exist for producing such cells. In order to develop novel stem cell protocols, we characterize at high resolution the development of the porcine medial entorhinal cortex by tracing neuronal and glial subtypes from mid-gestation to the adult brain to identify the transcriptomic profile of progenitor and adult stellate cells. Importantly, we could confirm the robustness of our data by extracting developmental factors from the identified intermediate stellate cell cluster and implemented these factors to generate putative intermediate stellate cells from human induced pluripotent stem cells. Six transcription factors identified from the stellate cell cluster including RUNX1T1, SOX5, FOXP1, MEF2C, TCF4, EYA2 were overexpressed using a forward programming approach to produce neurons expressing a unique combination of RELN, SATB2, LEF1 and BCL11B observed in stellate cells. Further analyses of the individual transcription factors led to the discovery that FOXP1 is critical in the reprogramming process and omission of RUNX1T1 and EYA2 enhances neuron conversion. Our findings contribute not only to the profiling of cell types within the developing and adult brain's medial entorhinal cortex but also provides proof-of-concept for using scRNAseq data to produce entorhinal intermediate stellate cells from human pluripotent stem cells in-vitro.

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.