Assessing parameter efficient methods for pre-trained language model in annotating scRNA-seq data

Annotating cell types of single-cell RNA sequencing (scRNA-seq) data is crucial for studying cellular heterogeneity in the tumor microenvironment. Recently, large-scale pre-trained language models (PLMs) have achieved significant progress in cell-type annotation of scRNA-seq data. This approach effectively addresses previous methods' shortcomings in performance and generalization. However, fine-tuning PLMs for different downstream tasks demands considerable computational resources, rendering it impractical. Hence, a new research branch introduces parameter-efficient fine-tuning (PEFT). This involves optimizing a few parameters while leaving the majority unchanged, leading to substantial reductions in computational expenses. Here, we utilize scBERT, a large-scale pre-trained model, to explore the capabilities of three PEFT methods in scRNA-seq cell type annotation. Extensive benchmark studies across several datasets demonstrate the superior applicability of PEFT methods. Furthermore, downstream analysis using models obtained through PEFT showcases their utility in novel cell type discovery and model interpretability for potential marker genes. Our findings underscore the considerable potential of PEFT in PLM-based cell type annotation, presenting novel perspectives for the analysis of scRNA-seq data.

scRNA-seq analysis of cells comprising the amphioxus notochord

Cephalochordates occupy a key phylogenetic position for deciphering the origin and evolution of chordates, since they diverged earlier than urochordates and vertebrates. The notochord is the most prominent feature of chordates. The amphioxus notochord features coin-shaped cells bearing myofibrils. Notochord-derived hedgehog signaling contributes to patterning of the dorsal nerve cord, as in vertebrates. However, properties of constituent notochord cells remain unknown at the single-cell level. We examined these properties using Iso-seq analysis, single-cell RNA-seq analysis, and in situ hybridization (ISH). Gene expression profiles broadly categorize notochordal cells into myofibrillar cells and non-myofibrillar cells. Myofibrillar cells occupy most of the central portion of the notochord, and some cells extend the notochordal horn to both sides of the ventral nerve cord. Some notochord myofibrillar genes are not expressed in myotomes, suggesting an occurrence of myofibrillar genes that are preferentially expressed in notochord. On the other hand, non-myofibrillar cells contain dorsal, lateral, and ventral Müller cells, and all three express both hedgehog and Brachyury. This was confirmed by ISH, although expression of hedgehog in ventral Müller cells was minimal. In addition, dorsal Müller cells express neural transmission-related genes, suggesting an interaction with nerve cord. Lateral Müller cells express hedgehog and other signaling-related genes, suggesting an interaction with myotomes positioned lateral to the notochord. Ventral Müller cells also expressed genes for FGF- and EGF-related signaling, which may be associated with development of endoderm, ventral to the notochord. Lateral Müller cells were intermediate between dorsal/ventral Müller cells. Since vertebrate notochord contributes to patterning and differentiation of ectoderm (nerve cord), mesoderm (somite), and endoderm, this investigation provides evidence that an ancestral or original form of vertebrate notochord is present in extant cephalochordates.

scRNA-seq analysis discovered suppression of immunomodulatory dependent inflammatory response in PMBCs exposed to silver nanoparticles

The assessment of AgNPs toxicity in vitro and in vivo models are frequently conflicting and inaccurate. Nevertheless, single cell immunological responses in a heterogenous environment have received little attention. Therefore, in this study, we have performed in-depth analysis which clearly revealed cellular-metal ion association as well as specific immunological response. Our study didn't show significant population differences in PMBC between control and AgNPs group implying no toxicological response. To confirm it further, deep profiling identified differences in subsets and differentially expressed genes (DEGs) of monocytes, B cells and T cells. Notably, monocyte subsets showed significant upregulation of metallothionein (MT) gene expression such as MT1G, MT1X, MT1E, MT1A, and MT1F. On the other hand, downregulation of pro-inflammatory genes such as IL1β and CCL3 in both CD16 + and CD16- monocyte subsets were observed. This result indicated that AgNPs association with monocyte subsets de-promoted inflammatory responsive genes suggesting no significant toxicity observed in AgNPs treated group. Other cell types such as B cells and T cells also showed negligible differences in their subsets suggesting no toxicity response. Further, AgNPs treated group showed upregulation of cell proliferation, ribosomal synthesis, downregulation of cytokine release, and T cell differentiation inhibition. Overall, our results conclude that treatment of AgNPs to PMBC cells didn't display immunological related cytotoxicity response and thus motivate researchers to use them actively for biomedical applications.

Single-cell transcriptomics identifies new blood cell populations in Drosophila released at the onset of metamorphosis

Drosophila blood cells called hemocytes form an efficient barrier against infections and tissue damage. During metamorphosis, hemocytes undergo tremendous changes in their shape and behavior, preparing them for tissue clearance. Yet, the diversity and functional plasticity of pupal blood cells have not been explored. Here, we combine single-cell transcriptomics and high-resolution microscopy to dissect the heterogeneity and plasticity of pupal hemocytes. We identified undifferentiated and specified hemocytes with different molecular signatures associated with distinct functions such as antimicrobial, antifungal immune defense, cell adhesion or secretion. Strikingly, we identified a highly migratory and immune-responsive pupal cell population expressing typical markers of the posterior signaling center (PSC), which is known to be an important niche in the larval lymph gland. PSC-like cells become restricted to the abdominal segments and are morphologically very distinct from typical Hemolectin (Hml)-positive plasmatocytes. G-TRACE lineage experiments further suggest that PSC-like cells can transdifferentiate to lamellocytes triggered by parasitoid wasp infestation. In summary, we present the first molecular description of pupal Drosophila blood cells, providing insights into blood cell functional diversification and plasticity during pupal metamorphosis.

Construction of a circRNA- lincRNA-lncRNA-miRNA-mRNA ceRNA regulatory network identifies genes and pathways linked to goat fertility

Background: There is growing interest in the genetic improvement of fertility traits in female goats. With high-throughput genotyping, single-cell RNA sequencing (scRNA-seq) is a powerful tool for measuring gene expression profiles. The primary objective was to investigate comparative transcriptome profiling of granulosa cells (GCs) of high- and low-fertility goats, using scRNA-seq. Methods: Thirty samples from Ji'ning Gray goats (n = 15 for high fertility and n = 15 for low fertility) were retrieved from publicly available scRNA-seq data. Functional enrichment analysis and a literature mining approach were applied to explore modules and hub genes related to fertility. Then, interactions between types of RNAs identified were predicted, and the ceRNA regulatory network was constructed by integrating these interactions with other gene regulatory networks (GRNs). Results and discussion: Comparative transcriptomics-related analyses identified 150 differentially expressed genes (DEGs) between high- and low-fertility groups, based on the fold change (≥5 and ≤-5) and false discovery rate (FDR <0.05). Among these genes, 80 were upregulated and 70 were downregulated. In addition, 81 mRNAs, 58 circRNAs, 8 lincRNAs, 19 lncRNAs, and 55 miRNAs were identified by literature mining. Furthermore, we identified 18 hub genes (SMAD1, SMAD2, SMAD3, SMAD4, TIMP1, ERBB2, BMP15, TGFB1, MAPK3, CTNNB1, BMPR2, AMHR2, TGFBR2, BMP4, ESR1, BMPR1B, AR, and TGFB2) involved in goat fertility. Identified biological networks and modules were mainly associated with ovary signature pathways. In addition, KEGG enrichment analysis identified regulating pluripotency of stem cells, cytokine-cytokine receptor interactions, ovarian steroidogenesis, oocyte meiosis, progesterone-mediated oocyte maturation, parathyroid and growth hormone synthesis, cortisol synthesis and secretion, and signaling pathways for prolactin, TGF-beta, Hippo, MAPK, PI3K-Akt, and FoxO. Functional annotation of identified DEGs implicated important biological pathways. These findings provided insights into the genetic basis of fertility in female goats and are an impetus to elucidate molecular ceRNA regulatory networks and functions of DEGs underlying ovarian follicular development.

A single-cell RNA-seq analysis of early larval cell-types of the starfish, Patiria pectinifera: Insights into evolution of the chordate body plan

Ambulacrarians (echinoderms and hemichordates) are a sister group to chordates; thus, their larval cell-types may provide clues about evolution of chordate body plans. Although most genic information accumulated to date pertains to sea urchin embryogenesis, starfish embryogenesis represents a more ancestral mode than that of sea urchins. We performed single-cell RNA-seq analysis of cell-types from gastrulae and bipinnarial larvae of the starfish, Patiria pectinifera, and categorized them into 22 clusters, each of which is composed of cells with specific, shared profiles of development-relevant gene expression. Oral and aboral ectoderm, apical plate, hindgut or archenteron, midgut or intestine, pharynx, endomesoderm, stomodeum, and mesenchyme of the gastrulae, and neurons, ciliary bands, enterocoel and muscle of larvae were characterized by expression profiles of at least two relevant transcription factor genes and signaling molecular genes. Expression of Hox2, Hox7, Hox9/10, and Hox11/13b was detected in cells of clusters that form the larval enterocoel. By comparing homologous gene expression profiles in chordate embryos, we discuss and propose how the chordate body plan evolved from a deuterostome ancestor, from which the echinoderm body plan also evolved.

scGMAAE: Gaussian mixture adversarial autoencoders for diversification analysis of scRNA-seq data

The progress of single-cell RNA sequencing (scRNA-seq) has led to a large number of scRNA-seq data, which are widely used in biomedical research. The noise in the raw data and tens of thousands of genes pose a challenge to capture the real structure and effective information of scRNA-seq data. Most of the existing single-cell analysis methods assume that the low-dimensional embedding of the raw data belongs to a Gaussian distribution or a low-dimensional nonlinear space without any prior information, which limits the flexibility and controllability of the model to a great extent. In addition, many existing methods need high computational cost, which makes them difficult to be used to deal with large-scale datasets. Here, we design and develop a depth generation model named Gaussian mixture adversarial autoencoders (scGMAAE), assuming that the low-dimensional embedding of different types of cells follows different Gaussian distributions, integrating Bayesian variational inference and adversarial training, as to give the interpretable latent representation of complex data and discover the statistical distribution of different types of cells. The scGMAAE is provided with good controllability, interpretability and scalability. Therefore, it can process large-scale datasets in a short time and give competitive results. scGMAAE outperforms existing methods in several ways, including dimensionality reduction visualization, cell clustering, differential expression analysis and batch effect removal. Importantly, compared with most deep learning methods, scGMAAE requires less iterations to generate the best results.

In Silico Stage-Matching of Human, Marmoset, Mouse, and Pig Embryos to Enhance Organ Development Through Interspecies Chimerism

Currently, there is a significant shortage of transplantable organs for patients in need. Interspecies chimerism and blastocyst complementation are alternatives for generating transplantable human organs in host animals such as pigs to meet this shortage. While successful interspecies chimerism and organ generation have been observed between evolutionarily close species such as rat and mouse, barriers still exist for more distant species pairs such as human-mouse, marmoset-mouse, human-pig, and others. One of the proposed barriers to chimerism is the difference in developmental stages between the donor cells and the host embryo at the time the cells are introduced into the host embryo. Hence, there is a logical effort to stage-match the donor cells with the host embryos for enhancing interspecies chimerism. In this study, we used an in silico approach to simultaneously stage-match the early developing embryos of four species, including human, marmoset, mouse, and pig based on transcriptome similarities. We used an unsupervised clustering algorithm to simultaneously stage-match all four species as well as Spearman's correlation analyses to stage-match pairs of donor-host species. From our stage-matching analyses, we found that the four stages that best matched with each other are the human blastocyst (E6/E7), the gastrulating mouse embryo (E6-E6.75), the marmoset late inner cell mass, and the pig late blastocyst. We further demonstrated that human pluripotent stem cells best matched with the mouse post-implantation stages. We also performed ontology analysis of the genes upregulated and commonly expressed between donor-host species pairs at their best matched stages. The stage-matching results predicted by this study will inform in vivo and in vitro interspecies chimerism and blastocyst complementation studies and can be used to match donor cells with host embryos between multiple species pairs to enhance chimerism for organogenesis.

Practical bioinformatics pipelines for single-cell RNA-seq data analysis

Single-cell RNA sequencing (scRNA-seq) is a revolutionary tool to explore cells. With an increasing number of scRNA-seq data analysis tools that have been developed, it is challenging for users to choose and compare their performance. Here, we present an overview of the workflow for computational analysis of scRNA-seq data. We detail the steps of a typical scRNA-seq analysis, including experimental design, pre-processing and quality control, feature selection, dimensionality reduction, cell clustering and annotation, and downstream analysis including batch correction, trajectory inference and cell-cell communication. We provide guidelines according to our best practice. This review will be helpful for the experimentalists interested in analyzing their data, and will aid the users seeking to update their analysis pipelines.

Comprehensive Analysis of Nasal Polyps Reveals a More Pronounced Type 2 Transcriptomic Profile of Epithelial Cells and Mast Cells in Aspirin-Exacerbated Respiratory Disease

Chronic rhinosinusitis with nasal polyps is affecting up to 3% of Western populations. About 10% of patients with nasal polyps also suffer from asthma and intolerance to aspirin, a syndrome called aspirin-exacerbated respiratory disease. Although eosinophilic inflammation is predominant in polyps of both diseases, phenotypic differences in the tissue-derived microenvironment, elucidating disease-specific characteristics, have not yet been identified. We sought to obtain detailed information about phenotypic and transcriptional differences in epithelial and immune cells in polyps of aspirin-tolerant and intolerant patients. Cytokine profiles in nasal secretions and serum of patients suffering from aspirin-exacerbated respiratory disease (n = 10) or chronic rhinosinusitis with nasal polyps (n = 9) were assessed using a multiplex mesoscale discovery assay. After enrichment for immune cell subsets by flow cytometry, we performed transcriptomic profiling by employing single-cell RNA sequencing. Aspirin-intolerant patients displayed significantly elevated IL-5 and CCL17 levels in nasal secretions corresponding to a more pronounced eosinophilic type 2 inflammation. Transcriptomic profiling revealed that epithelial and mast cells not only complement one another in terms of gene expression associated with the 15-lipoxygenase pathway but also show a clear type 2-associated inflammatory phenotype as identified by the upregulation of POSTN, CCL26, and IL13 in patients with aspirin-exacerbated respiratory disease. Interestingly, we also observed cellular stress responses indicated by an increase of MTRNR2L12, MTRNR2L8, and NEAT1 across all immune cell subsets in this disease entity. In conclusion, our findings support the hypothesis that epithelial and mast cells act in concert as potential drivers of the pathogenesis of the aspirin-exacerbated respiratory disease.

A Preliminary Single-Cell RNA-Seq Analysis of Embryonic Cells That Express Brachyury in the Amphioxus, Branchiostoma japonicum

Among chordate taxa, the cephalochordates diverged earlier than urochordates and vertebrates; thus, they retain unique, primitive developmental features. In particular, the amphioxus notochord has muscle-like properties, a feature not seen in urochordates or vertebrates. Amphioxus contains two Brachyury genes, Bra1 and Bra2. Bra2 is reportedly expressed in the blastopore, notochord, somites, and tail bud, in contrast to a low level of Bra1 expression only in notochord. To distinguish the expression profiles of the two Brachyury genes at the single-cell level, we carried out single-cell RNA-seq (scRNA-seq) analysis using the amphioxus, Branchiostoma japonicum. This scRNA-seq analysis classified B. japonicum embryonic cells into 15 clusters at developmental stages from midgastrula to early swimming larva. Brachyury was expressed in cells of clusters 4, 5, 8, and 9. We first confirmed that cluster 8 comprises cells that form somites since this cluster specifically expresses four myogenic factor genes. Cluster 9 contains a larger number of cells with high levels of Bra2 expression and a smaller number of cells with Bra1 expression. Simultaneous expression in cluster 9 of tool-kit genes, including FoxA, Goosecoid, and hedgehog, showed that this cluster comprises cells that form the notochord. Expression of Bra2, but not Bra1, in cells of clusters 4 and 5 at the gastrula stage together with expression of Wnt1 and Caudal indicates that clusters 4 and 5 comprise cells of the blastopore, which contiguously form the tail bud. In addition, Hox1, Hox3, and Hox4 were highly expressed in Bra2-expressing clusters 4, 5, 8, and 9 in a temporally coordinated manner, suggesting roles of anterior Hox genes in specification of mesodermal organs, including somites, notochord, and tail bud. This scRNA-seq analysis therefore highlights differences between the two Brachyury genes in relation to embryonic regions in which they are expressed and their levels of expression. Bra2 is the ancestral Brachyury in amphioxus, since expression in the blastopore is shared with other deuterostomes. On the other hand, Bra1 is a duplicate copy and likely evolved a supplementary function in notochord and somite formation in the Branchiostoma lineage.

Single Cell Transcriptome Data Analysis Defines the Heterogeneity of Peripheral Nerve Cells in Homeostasis and Regeneration

The advances in single-cell RNA sequencing technologies and the development of bioinformatics pipelines enable us to more accurately define the heterogeneity of cell types in a selected tissue. In this report, we re-analyzed recently published single-cell RNA sequencing data sets and provide a rationale to redefine the heterogeneity of cells in both intact and injured mouse peripheral nerves. Our analysis showed that, in both intact and injured peripheral nerves, cells could be functionally classified into four categories: Schwann cells, nerve fibroblasts, immune cells, and cells associated with blood vessels. Nerve fibroblasts could be sub-clustered into epineurial, perineurial, and endoneurial fibroblasts. Identified immune cell clusters include macrophages, mast cells, natural killer cells, T and B lymphocytes as well as an unreported cluster of neutrophils. Cells associated with blood vessels include endothelial cells, vascular smooth muscle cells, and pericytes. We show that endothelial cells in the intact mouse sciatic nerve have three sub-types: epineurial, endoneurial, and lymphatic endothelial cells. Analysis of cell type-specific gene changes revealed that Schwann cells and endoneurial fibroblasts are the two most important cell types promoting peripheral nerve regeneration. Analysis of communication between these cells identified potential signals for early blood vessel regeneration, neutrophil recruitment of macrophages, and macrophages activating Schwann cells. Through this analysis, we also report appropriate marker genes for future single cell transcriptome data analysis to identify cell types in intact and injured peripheral nerves. The findings from our analysis could facilitate a better understanding of cell biology of peripheral nerves in homeostasis, regeneration, and disease.

Implications of liver injury in risk-stratification and management of patients with COVID-19

BACKGROUND

Infection with SARS-CoV-2 has been associated with liver dysfunction, aggravation of liver burden, and liver injury. This study aimed to assess the effects of liver injuries on the clinical outcomes of patients with COVID-19.

METHODS

A total of 1520 patients with severe or critical COVID-19 from Huoshenshan Hospital, Wuhan, were enrolled. Chronic liver disease (CLD) was confirmed by consensus diagnostic criteria. Laboratory test results were compared between different groups. scRNA-seq data and bulk gene expression profiles were used to identify cell types associated with liver injury.

RESULTS

A total of 10.98% of patients with severe or critical COVID-19 developed liver injury after admission that was associated with significantly higher rates of mortality (21.74%, p < 0.001) and intensive care unit admission (26.71%, p < 0.001). Pre-existing CLDs were not associated with a higher risk. However, fatty liver disease and cirrhosis were associated with higher risks, supported by evidences from single cell and bulk transcriptome analysis that showed more TMPRSS2⁺ cells in these tissues. By generating a model, we were able to predict the risk and severity of liver injury during hospitalization.

CONCLUSION

We demonstrate that liver injury occurring during therapy as well as pre-existing CLDs like fatty liver disease and cirrhosis in patients with COVID-19 is significantly associated with the severity of disease and mortality, but the presence of other CLD is not associated. We provide a risk-score model that can predict whether patients with COVID-19 will develop liver injury or proceed to higher-risk stages during subsequent hospitalizations.