Metric multidimensional scaling for large single-cell datasets using neural networks.

Metric multidimensional scaling is one of the classical methods for embedding data into low-dimensional Euclidean space. It creates the low-dimensional embedding by approximately preserving the pairwise distances between the input points. However, current state-of-the-art approaches only scale to a few thousand data points. For larger data sets such as those occurring in single-cell RNA sequencing experiments, the running time becomes prohibitively large and thus alternative methods such as PCA are widely used instead. Here, we propose a simple neural network-based approach for solving the metric multidimensional scaling problem that is orders of magnitude faster than previous state-of-the-art approaches, and hence scales to data sets with up to a few million cells. At the same time, it provides a non-linear mapping between high- and low-dimensional space that can place previously unseen cells in the same embedding.

Combinatorial microRNA activity is essential for the transition of pluripotent cells from proliferation into dormancy.

Dormancy is a key feature of stem cell function in adult tissues as well as in embryonic cells in the context of diapause. The establishment of dormancy is an active process that involves extensive transcriptional, epigenetic, and metabolic rewiring. How these processes are coordinated to successfully transition cells to the resting dormant state remains unclear. Here we show that microRNA activity, which is otherwise dispensable for preimplantation development, is essential for the adaptation of early mouse embryos to the dormant state of diapause. In particular, the pluripotent epiblast depends on miRNA activity, the absence of which results in the loss of pluripotent cells. Through the integration of high-sensitivity small RNA expression profiling of individual embryos and protein expression of miRNA targets with public data of protein-protein interactions, we constructed the miRNA-mediated regulatory network of mouse early embryos specific to diapause. We find that individual miRNAs contribute to the combinatorial regulation by the network, and the perturbation of the network compromises embryo survival in diapause. We further identified the nutrient-sensitive transcription factor TFE3 as an upstream regulator of diapause-specific miRNAs, linking cytoplasmic MTOR activity to nuclear miRNA biogenesis. Our results place miRNAs as a critical regulatory layer for the molecular rewiring of early embryos to establish dormancy.

Transcript Isoform Diversity of Y Chromosome Ampliconic Genes of Great Apes Uncovered Using Long Reads and Telomere-to-Telomere Reference Genome Assemblies.

Y chromosomes of great apes harbor Ampliconic Genes (YAGs)-multi-copy gene families (BPY2, CDY, DAZ, HSFY, PRY, RBMY, TSPY, VCY, and XKRY) that encode proteins important for spermatogenesis. Previous work assembled YAG transcripts based on their targeted sequencing but not using reference genome assemblies, potentially resulting in an incomplete transcript repertoire. Here we used the recently produced gapless telomere-to-telomere (T2T) Y chromosome assemblies of great ape species (bonobo, chimpanzee, human, gorilla, Bornean orangutan, and Sumatran orangutan) and analyzed RNA data from whole-testis samples for the same species. We generated hybrid transcriptome assemblies by combining targeted long reads (Pacific Biosciences), untargeted long reads (Pacific Biosciences) and untargeted short reads (Illumina)and mapping them to the T2T reference genomes. Compared to the results from the reference-free approach, average transcript length was more than two times higher, and the total number of transcripts decreased three times, improving the quality of the assembled transcriptome. The reference-based transcriptome assemblies allowed us to differentiate transcripts originating from different Y chromosome gene copies and from their non-Y chromosome homologs. We identified two sources of transcriptome diversity-alternative splicing and gene duplication with subsequent diversification of gene copies. For each gene family, we detected transcribed pseudogenes along with protein-coding gene copies. We revealed previously unannotated gene copies of YAGs as compared to currently available NCBI annotations, as well as novel isoforms for annotated gene copies. This analysis paves the way for better understanding Y chromosome gene functions, which is important given their role in spermatogenesis.

Correcting PCR amplification errors in unique molecular identifiers to generate accurate numbers of sequencing molecules.

Unique molecular identifiers are random oligonucleotide sequences that remove PCR amplification biases. However, the impact that PCR associated sequencing errors have on the accuracy of generating absolute counts of RNA molecules is underappreciated. We show that PCR errors are a source of inaccuracy in both bulk and single-cell sequencing data, and synthesizing unique molecular identifiers using homotrimeric nucleotide blocks provides an error-correcting solution that allows absolute counting of sequenced molecules.

Counting pseudoalignments to novel splicing events.

MOTIVATION: Alternative splicing (AS) of introns from pre-mRNA produces diverse sets of transcripts across cell types and tissues, but is also dysregulated in many diseases. Alignment-free computational methods have greatly accelerated the quantification of mRNA transcripts from short RNA-seq reads, but they inherently rely on a catalog of known transcripts and might miss novel, disease-specific splicing events. By contrast, alignment of reads to the genome can effectively identify novel exonic segments and introns. Event-based methods then count how many reads align to predefined features. However, an alignment is more expensive to compute and constitutes a bottleneck in many AS analysis methods. RESULTS: Here, we propose fortuna, a method that guesses novel combinations of annotated splice sites to create transcript fragments. It then pseudoaligns reads to fragments using kallisto and efficiently derives counts of the most elementary splicing units from kallisto's equivalence classes. These counts can be directly used for AS analysis or summarized to larger units as used by other widely applied methods. In experiments on synthetic and real data, fortuna was around 7× faster than traditional align and count approaches, and was able to analyze almost 300 million reads in just 15 min when using four threads. It mapped reads containing mismatches more accurately across novel junctions and found more reads supporting aberrant splicing events in patients with autism spectrum disorder than existing methods. We further used fortuna to identify novel, tissue-specific splicing events in Drosophila. AVAILABILITY AND IMPLEMENTATION: fortuna source code is available at https://github.com/canzarlab/fortuna.

The function of Wtap in N6-adenosine methylation of mRNAs controls T cell receptor signaling and survival of T cells.

T cell antigen-receptor (TCR) signaling controls the development, activation and survival of T cells by involving several layers and numerous mechanisms of gene regulation. N6-methyladenosine (m6A) is the most prevalent messenger RNA modification affecting splicing, translation and stability of transcripts. In the present study, we describe the Wtap protein as essential for m6A methyltransferase complex function and reveal its crucial role in TCR signaling in mouse T cells. Wtap and m6A methyltransferase functions were required for the differentiation of thymocytes, control of activation-induced death of peripheral T cells and prevention of colitis by enabling gut RORγt+ regulatory T cell function. Transcriptome and epitranscriptomic analyses reveal that m6A modification destabilizes Orai1 and Ripk1 mRNAs. Lack of post-transcriptional repression of the encoded proteins correlated with increased store-operated calcium entry activity and diminished survival of T cells with conditional genetic inactivation of Wtap. These findings uncover how m6A modification impacts on TCR signal transduction and determines activation and survival of T cells.

Ccr4-Not complex reduces transcription efficiency in heterochromatin.

Heterochromatic silencing is thought to occur through a combination of transcriptional silencing and RNA degradation, but the relative contribution of each pathway is not known. In this study, we analyzed RNA Polymerase II (RNA Pol II) occupancy and levels of nascent and steady-state RNA in different mutants of Schizosaccharomyces pombe, in order to quantify the contribution of each pathway to heterochromatic silencing. We found that transcriptional silencing consists of two components, reduced RNA Pol II accessibility and, unexpectedly, reduced transcriptional efficiency. Heterochromatic loci showed lower transcriptional output compared to euchromatic loci, even when comparable amounts of RNA Pol II were present in both types of regions. We determined that the Ccr4-Not complex and H3K9 methylation are required for reduced transcriptional efficiency in heterochromatin and that a subset of heterochromatic RNA is degraded more rapidly than euchromatic RNA. Finally, we quantified the contribution of different chromatin modifiers, RNAi and RNA degradation to each silencing pathway. Our data show that several pathways contribute to heterochromatic silencing in a locus-specific manner and reveal transcriptional efficiency as a new mechanism of silencing.

Genome-wide off-target analyses of CRISPR/Cas9-mediated T-cell receptor engineering in primary human T cells.

OBJECTIVES: Exploiting the forces of human T cells for treatment has led to the current paradigm of emerging immunotherapy strategies. Genetic engineering of the T-cell receptor (TCR) redirects specificity, ablates alloreactivity and brings significant progress and off-the-shelf options to emerging adoptive T-cell transfer (ACT) approaches. Targeted CRISPR/Cas9-mediated double-strand breaks in the DNA enable knockout or knock-in engineering. METHODS: Here, we perform CRISPR/Cas9-mediated TCR knockout using a therapeutically relevant ribonucleoprotein (RNP) delivery method to assess the safety of genetically engineered T-cell products. Whole-genome sequencing was performed to analyse whether CRISPR/Cas9-mediated DNA double-strand break at the TCR locus is associated with off-target events in human primary T cells. RESULTS: TCRα chain and TCRß chain knockout leads to high on-target InDel frequency and functional knockout. None of the predicted off-target sites could be confirmed experimentally, whereas whole-genome sequencing and manual Integrative Genomics Viewer (IGV) review revealed 9 potential low-frequency off-target events genome-wide. Subsequent amplification and targeted deep sequencing in 7 of 7 evaluable loci did not confirm these low-frequency InDels. Therefore, off-target events are unlikely to be caused by the CRISPR/Cas9 engineering. CONCLUSION: The combinatorial approach of whole-genome sequencing and targeted deep sequencing confirmed highly specific genetic engineering using CRISPR/Cas9-mediated TCR knockout without potentially harmful exonic off-target effects.

Partitioning RNAs by length improves transcriptome reconstruction from short-read RNA-seq data.

The accuracy of methods for assembling transcripts from short-read RNA sequencing data is limited by the lack of long-range information. Here we introduce Ladder-seq, an approach that separates transcripts according to their lengths before sequencing and uses the additional information to improve the quantification and assembly of transcripts. Using simulated data, we show that a kallisto algorithm extended to process Ladder-seq data quantifies transcripts of complex genes with substantially higher accuracy than conventional kallisto. For reference-based assembly, a tailored scheme based on the StringTie2 algorithm reconstructs a single transcript with 30.8% higher precision than its conventional counterpart and is more than 30% more sensitive for complex genes. For de novo assembly, a similar scheme based on the Trinity algorithm correctly assembles 78% more transcripts than conventional Trinity while improving precision by 78%. In experimental data, Ladder-seq reveals 40% more genes harboring isoform switches compared to conventional RNA sequencing and unveils widespread changes in isoform usage upon m6A depletion by Mettl14 knockout.

CYFIP1 Dosages Exhibit Divergent Behavioral Impact via Diametric Regulation of NMDA Receptor Complex Translation in Mouse Models of Psychiatric Disorders.

BACKGROUND: Gene dosage imbalance caused by copy number variations (CNVs) is a prominent contributor to brain disorders. In particular, 15q11.2 CNV duplications and deletions have been associated with autism spectrum disorder and schizophrenia, respectively. The mechanism underlying these diametric contributions remains unclear. METHODS: We established both loss-of-function and gain-of-function mouse models of Cyfip1, one of four genes within 15q11.2 CNVs. To assess the functional consequences of altered CYFIP1 levels, we performed systematic investigations on behavioral, electrophysiological, and biochemical phenotypes in both mouse models. In addition, we utilized RNA immunoprecipitation sequencing (RIP-seq) analysis to reveal molecular targets of CYFIP1 in vivo. RESULTS: Cyfip1 loss-of-function and gain-of function mouse models exhibited distinct and shared behavioral abnormalities related to autism spectrum disorder and schizophrenia. RIP-seq analysis identified messenger RNA targets of CYFIP1 in vivo, including postsynaptic NMDA receptor (NMDAR) complex components. In addition, these mouse models showed diametric changes in levels of postsynaptic NMDAR complex components at synapses because of dysregulated protein translation, resulting in bidirectional alteration of NMDAR-mediated signaling. Importantly, pharmacological balancing of NMDAR signaling in these mouse models with diametric Cyfip1 dosages rescues behavioral abnormalities. CONCLUSIONS: CYFIP1 regulates protein translation of NMDAR and associated complex components at synapses to maintain normal synaptic functions and behaviors. Our integrated analyses provide insight into how gene dosage imbalance caused by CNVs may contribute to divergent neuropsychiatric disorders.

A generalization of t-SNE and UMAP to single-cell multimodal omics.

Emerging single-cell technologies profile multiple types of molecules within individual cells. A fundamental step in the analysis of the produced high-dimensional data is their visualization using dimensionality reduction techniques such as t-SNE and UMAP. We introduce j-SNE and j-UMAP as their natural generalizations to the joint visualization of multimodal omics data. Our approach automatically learns the relative contribution of each modality to a concise representation of cellular identity that promotes discriminative features but suppresses noise. On eight datasets, j-SNE and j-UMAP produce unified embeddings that better agree with known cell types and that harmonize RNA and protein velocity landscapes.

Linear-time cluster ensembles of large-scale single-cell RNA-seq and multimodal data.

A fundamental task in single-cell RNA-seq (scRNA-seq) analysis is the identification of transcriptionally distinct groups of cells. Numerous methods have been proposed for this problem, with a recent focus on methods for the cluster analysis of ultralarge scRNA-seq data sets produced by droplet-based sequencing technologies. Most existing methods rely on a sampling step to bridge the gap between algorithm scalability and volume of the data. Ignoring large parts of the data, however, often yields inaccurate groupings of cells and risks overlooking rare cell types. We propose method Specter that adopts and extends recent algorithmic advances in (fast) spectral clustering. In contrast to methods that cluster a (random) subsample of the data, we adopt the idea of landmarks that are used to create a sparse representation of the full data from which a spectral embedding can then be computed in linear time. We exploit Specter's speed in a cluster ensemble scheme that achieves a substantial improvement in accuracy over existing methods and identifies rare cell types with high sensitivity. Its linear-time complexity allows Specter to scale to millions of cells and leads to fast computation times in practice. Furthermore, on CITE-seq data that simultaneously measures gene and protein marker expression, we show that Specter is able to use multimodal omics measurements to resolve subtle transcriptomic differences between subpopulations of cells.

McSplicer: a probabilistic model for estimating splice site usage from RNA-seq data.

MOTIVATION: Alternative splicing removes intronic sequences from pre-mRNAs in alternative ways to produce different forms (isoforms) of mature mRNA. The composition of expressed transcripts gives specific functionalities to cells in a particular condition or developmental stage. In addition, a large fraction of human disease mutations affect splicing and lead to aberrant mRNA and protein products. Current methods that interrogate the transcriptome based on RNA-seq either suffer from short read length when trying to infer full-length transcripts, or are restricted to predefined units of alternative splicing that they quantify from local read evidence. RESULTS: Instead of attempting to quantify individual outcomes of the splicing process such as local splicing events or full-length transcripts, we propose to quantify alternative splicing using a simplified probabilistic model of the underlying splicing process. Our model is based on the usage of individual splice sites and can generate arbitrarily complex types of splicing patterns. In our implementation, McSplicer, we estimate the parameters of our model using all read data at once and we demonstrate in our experiments that this yields more accurate estimates compared to competing methods. Our model is able to describe multiple effects of splicing mutations using few, easy to interpret parameters, as we illustrate in an experiment on RNA-seq data from autism spectrum disorder patients. AVAILABILITY: McSplicer source code is available at https://github.com/canzarlab/McSplicer and has been deposited in archived format at https://doi.org/10.5281/zenodo.4449881. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BiCoN: network-constrained biclustering of patients and omics data.

MOTIVATION: Unsupervised learning approaches are frequently used to stratify patients into clinically relevant subgroups and to identify biomarkers such as disease-associated genes. However, clustering and biclustering techniques are oblivious to the functional relationship of genes and are thus not ideally suited to pinpoint molecular mechanisms along with patient subgroups. RESULTS: We developed the network-constrained biclustering approach Biclustering Constrained by Networks (BiCoN) which (i) restricts biclusters to functionally related genes connected in molecular interaction networks and (ii) maximizes the difference in gene expression between two subgroups of patients. This allows BiCoN to simultaneously pinpoint molecular mechanisms responsible for the patient grouping. Network-constrained clustering of genes makes BiCoN more robust to noise and batch effects than typical clustering and biclustering methods. BiCoN can faithfully reproduce known disease subtypes as well as novel, clinically relevant patient subgroups, as we could demonstrate using breast and lung cancer datasets. In summary, BiCoN is a novel systems medicine tool that combines several heuristic optimization strategies for robust disease mechanism extraction. BiCoN is well-documented and freely available as a python package or a web interface. AVAILABILITY AND IMPLEMENTATION: PyPI package: https://pypi.org/project/bicon. WEB INTERFACE: https://exbio.wzw.tum.de/bicon. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Validation strategies for antibodies targeting modified ribonucleotides.

Chemical modifications are found on almost all RNAs and affect their coding and noncoding functions. The identification of m6A on mRNA and its important role in gene regulation stimulated the field to investigate whether additional modifications are present on mRNAs. Indeed, modifications including m1A, m5C, m7G, 2'-OMe, and Ψ were detected. However, since their abundances are low and tools used for their corroboration are often not well characterized, their physiological relevance remains largely elusive. Antibodies targeting modified nucleotides are often used but have limitations such as low affinity or specificity. Moreover, they are not always well characterized and due to the low abundance of the modification, particularly on mRNAs, generated data sets might resemble noise rather than specific modification patterns. Therefore, it is critical that the affinity and specificity is rigorously tested using complementary approaches. Here, we provide an experimental toolbox that allows for testing antibody performance prior to their use.

Sphetcher: Spherical Thresholding Improves Sketching of Single-Cell Transcriptomic Heterogeneity.

The massive size of single-cell RNA sequencing datasets often exceeds the capability of current computational analysis methods to solve routine tasks such as detection of cell types. Recently, geometric sketching was introduced as an alternative to uniform subsampling. It selects a subset of cells (the sketch) that evenly cover the transcriptomic space occupied by the original dataset, to accelerate downstream analyses and highlight rare cell types. Here, we propose algorithm Sphetcher that makes use of the thresholding technique to efficiently pick representative cells within spheres (as opposed to the typically used equal-sized boxes) that cover the entire transcriptomic space. We show that the spherical sketch computed by Sphetcher constitutes a more accurate representation of the original transcriptomic landscape. Our optimization scheme allows to include fairness aspects that can encode prior biological or experimental knowledge. We show how a fair sampling can inform the inference of the trajectory of human skeletal muscle myoblast differentiation.

Clinical presentation and differential splicing of SRSF2, U2AF1 and SF3B1 mutations in patients with acute myeloid leukemia.

Previous studies demonstrated that splicing factor mutations are recurrent events in hematopoietic malignancies with both clinical and functional implications. However, their aberrant splicing patterns in acute myeloid leukemia remain largely unexplored. In this study, we characterized mutations in SRSF2, U2AF1, and SF3B1, the most commonly mutated splicing factors. In our clinical analysis of 2678 patients, splicing factor mutations showed inferior relapse-free and overall survival, however, these mutations did not represent independent prognostic markers. RNA-sequencing of 246 and independent validation in 177 patients revealed an isoform expression profile which is highly characteristic for each individual mutation, with several isoforms showing a strong dysregulation. By establishing a custom differential splice junction usage pipeline, we accurately detected aberrant splicing in splicing factor mutated samples. A large proportion of differentially used junctions were novel, including several junctions in leukemia-associated genes. In SRSF2(P95H) mutants, we further explored the possibility of a cascading effect through the dysregulation of the splicing pathway. Furthermore, we observed a validated impact on overall survival for two junctions overused in SRSF2(P95H) mutants. We conclude that splicing factor mutations do not represent independent prognostic markers. However, they do have genome-wide consequences on gene splicing leading to dysregulated isoform expression of several genes.

Leukemia-induced dysfunctional TIM-3+CD4+ bone marrow T cells increase risk of relapse in pediatric B-precursor ALL patients.

Interaction of malignancies with tissue-specific immune cells has gained interest for prognosis and intervention of emerging immunotherapies. We analyzed bone marrow T cells (bmT) as tumor-infiltrating lymphocytes in pediatric precursor-B cell acute lymphoblastic leukemia (ALL). Based on data from 100 patients, we show that ALL is associated with late-stage CD4+ phenotype and loss of early CD8+ T cells. The inhibitory exhaustion marker TIM-3 on CD4+ bmT increased relapse risk (RFS = 94.6/70.3%) confirmed by multivariate analysis. The hazard ratio of TIM-3 expression nearly reached the hazard ratio of MRD (7.1 vs. 8.0) indicating that patients with a high frequency of TIM-3+CD4+ bone marrow T cells at initial diagnosis have a 7.1-fold increased risk to develop ALL relapse. Comparison of wild type primary T cells to CRISPR/Cas9-mediated TIM-3 knockout and TIM-3 overexpression confirmed the negative effect of TIM-3 on T cell responses against ALL. TIM-3+CD4+ bmT are increased in ALL overexpressing CD200, that leads to dysfunctional antileukemic T cell responses. In conclusion, TIM-3-mediated interaction between bmT and leukemia cells is shown as a strong risk factor for relapse in pediatric B-lineage ALL. CD200/TIM-3-signaling, rather than PD-1/PD-L1, is uncovered as a mechanism of T cell dysfunction in ALL with major implication for future immunotherapies.

Chromatyping: Reconstructing Nucleosome Profiles from NOMe Sequencing Data.

Measuring nucleosome positioning in cells is crucial for the analysis of epigenetic gene regulation. Reconstruction of nucleosome profiles of individual cells or subpopulations of cells remains challenging because most genome-wide assays measure nucleosome positioning and DNA accessibility for thousands of cells using bulk sequencing. In this study we use characteristics of the NOMe (nucleosome occupancy and methylation)-sequencing assay to derive a new approach, called ChromaClique, for deconvolution of different nucleosome profiles (chromatypes) from cell subpopulations of one NOMe-seq measurement. ChromaClique uses a maximal clique enumeration algorithm on a newly defined NOMe read graph that is able to group reads according to their nucleosome profiles. We show that the edge probabilities of that graph can be efficiently computed using hidden Markov models. We demonstrate using simulated data that ChromaClique is more accurate than a related method and scales favorably, allowing genome-wide analyses of chromatypes in cell subpopulations.

Alternative splicing regulates stochastic NLRP3 activity.

Leucine-rich repeat (LRR) domains are evolutionarily conserved in proteins that function in development and immunity. Here we report strict exonic modularity of LRR domains of several human gene families, which is a precondition for alternative splicing (AS). We provide evidence for AS of LRR domain within several Nod-like receptors, most prominently the inflammasome sensor NLRP3. Human NLRP3, but not mouse NLRP3, is expressed as two major isoforms, the full-length variant and a variant lacking exon 5. Moreover, NLRP3 AS is stochastically regulated, with NLRP3 ∆ exon 5 lacking the interaction surface for NEK7 and hence loss of activity. Our data thus reveals unexpected regulatory roles of AS through differential utilization of LRRs modules in vertebrate innate immunity.