The transcriptome, the complete set of RNA molecules within a cell, plays a critical role in regulating physiological processes. The advent of RNA sequencing (RNA-seq) facilitated by Next Generation Sequencing (NGS) technologies, has revolutionized transcriptome research, providing unique insights into gene expression dynamics. This powerful strategy can be applied at both bulk tissue and single-cell levels. Bulk RNA-seq provides a gene expression profile within a tissue sample. Conversely, single-cell RNA sequencing (scRNA-seq) offers resolution at the cellular level, allowing the uncovering of cellular heterogeneity, identification of rare cell types, and distinction between distinct cell populations. As computational tools, machine learning techniques, and NGS sequencing platforms continue to evolve, the field of transcriptome research is poised for significant advancements. Therefore, to fully harness this potential, a comprehensive understanding of bulk RNA-seq and scRNA-seq technologies, including their advantages, limitations, and computational considerations, is crucial. This review provides a systematic comparison of the computational processes involved in both RNA-seq and scRNA-seq, highlighting their fundamental principles, applications, strengths, and limitations, while outlining future directions in transcriptome research.
It is increasingly clear that cancer is a complex systemic disease and one of the most fatal diseases in humans. Complex systems, including cancer, exhibit critical transitions in which the system abruptly shifts from one state to another. However, predicting these critical transitions is difficult as the system may show little change before the tipping point is reached. Models for predicting cancer are generally not accurate enough to reliably predict where these critical transitions will occur. Additionally, there is often a gap between theoretical results and clinical practice. To address these issues, we conducted a study using gastric cancer as a representative to reveal the tipping point of cancer and develop a feasible method for clinical monitoring. We used gene regulatory networks and a landscape framework to quantify the formation of gastric cancer. Since the dissipation cost of cancer cells is different from that of normal cells, we calculated the entropy product rate (EPR) and mean flux to quantify the thermodynamic cost and dynamical driving force in predicting critical transitions of cancer, which can serve as early warning signals. Both the EPR and mean flux change sharply near the point when the cancer state is about to emerge and/or the normal state is about to disappear. Moreover, the peak or sharp upward trends of the signals occur much earlier than critical slowdown and flickering frequency. These significant variations can be used as early warning signals for cancer. To further explore early warning signals in clinical and experimental trials, we calculated the difference in cross correlations (ΔC) forward and backward in time for the stochastic gene expression time series. This time-irreversible measure gives a rise to peak before the bifurcation points, which can help detect precancerous and metastatic early warning signals in clinical practice rather than just theoretical calculation. This study is crucial for effectively identifying early warning signals for cancer in clinical and experimental settings.
Iron is a critical yet limited nutrient for microbial growth. To scavenge iron, most microbes produce siderophores—diverse small molecules with high iron affinities. Different siderophores are specifically recognized and uptaken by corresponding recognizers, enabling targeted interventions and intriguing cheater-producer dynamics. We propose constructing a comprehensive iron interaction network, or “iron-net”, across the microbial world. Such a network offers the potentialfor precise manipulation of the microbiota, with conceivable applications in medicine, agriculture, and industry as well as advancing microbialecologyandevolutiontheories.Previously,oursuccessfulconstruction of an iron-net in the Pseudomonas genus demonstrated the feasibility of coevolution-inspired digital siderophore-typing. Enhanced by machine learning techniques and expanding sequencing data, forging such an iron-net calls for multidisciplinary collaborations and holds significant promise in addressing critical challenges in microbial communities.
Cell clustering plays a pivotal role in deciphering the intricacies of cell types, facilitating subsequent cell annotation endeavors within scRNA-seq data analysis. In this paper, we propose a novel swapped contrastive clustering algorithm for scRNA-seq data called scSCC. scSCC combines two contrastive learning modules, namely the instance contrastive learning module and the swapped prediction module, to extract clustering-friendly cell representations. Through the combination of swapped prediction module and instance contrastive learning module, scSCC can retrieve disentangled cell representations and amplify the clustering signals in the latent space, leading to satisfactory clustering performance. Different from existing contrastive-learning-based scRNA-seq data clustering algorithms, the swapped prediction module of scSCC injects clustering signals to the latent space through some clustering prototypes. The swapped prediction module encourages cells of the same cluster to gravitate toward the common clustering prototype and naturally stay away from other prototypes in the latent space, hence cell representations obtained by scSCC are more clustering-friendly compared to other algorithms. Experimental results on real scRNA-seq datasets show that scSCC achieves improved clustering performance compared with the benchmark methods. The ablation study on two contrastive modules exhibits the promotion by the combination of instance learning module and swapped prediction module. The source codes are available at the GitHub website (EnchantedJoy/scSCC).
Abscisic acid (ABA)-responsive elements (ABREs) are the major cis-regulatory elements in ABA-induced gene expression. However, the impact of sequence variations on ABRE function is not yet well-understood. Here, we used synthetic STARR-seq to quantitatively assess the effects of single-nucleotide substitutions on ABRE activity. Our results revealed that the nucleotide substitutions in both the ACGT-core and ACGT-flank regions affected transcriptional strength. Alterations in the ACGT-core sequence had a more significant impact on ABRE activity than changes in the flanking region. Interestingly, we observed that the ACGT-flank variants with high activity exhibited a strong sequence preference in the downstream region, whereas the highly active core variants were diverse in sequence patterns. Our studies provide a quantitative map of ABRE activity at single-nucleotide resolution, which will facilitate the design of ABA-responsive promoters with desired activities in plants.
The divergence rate between the alignable genomes of humans and chimpanzees is as little as 1.23%. Their phenotypical difference was hypothesized to be accounted for by gene regulation. We construct the cis-regulatory element frequency (CREF) matrix to represent the proximal regulatory sequences for each species. Each CREF matrix is further decomposed into dual eigen-modules. By comparing the CREF modules of four existing hominid species, we examine their quantitative and qualitative changes along evolution. We identified two saltations: one between the 4th and 5th, the other between the 9th and 10th eigen-levels. The cognition and intelligence unique to humans are thus found from the saltations at the molecular level. They include long-term memory, cochlea/inner ear morphogenesis that enables the development of human language/music, social behavior that allows us to live together peacefully and to work collaboratively, and visual/observational/associative learning. Moreover, we found exploratory behavior crucial for humans’ creativity, the GABA-B receptor activation that protects our neurons, and serotonin biosynthesis/signaling that regulates our happiness. We observed a remarkable increase in the number of motifs present on Alu elements on the 4th/9th motif-eigenvectors. The cognition and intelligence unique to humans can, by and large, be identified using only the CREF profiles without any a priori. Although gradual evolution might be the only mode in the mutations of protein sequences, the evolution of gene regulation has both gradual and saltational modes, which could be explained by the framework of CREF eigen-modules.
Asymmetry between outer and inner leaflets of cell membrane, such as variations in phospholipid composition, cholesterol (CHOL) distribution, stress levels, and ion environments, could significantly influence the biophysical properties of membranes, including the lateral organization of lipids and the formation of membrane nanodomains. To elucidate the effects of lipid component, lipid number mismatch, CHOL concentration asymmetry, and ionic conditions on membrane properties, we constructed several sets of all-atom, multi-component lipid bilayer models. Using molecular dynamics (MD) simulations, we investigated how membrane asymmetry modulates its biological characteristics. Our results indicate that CHOL concentration, whether symmetric or asymmetric between the leaflets, is the primary factor affecting membrane thickness, order parameters of the lipid tail, tilting angles of lipid molecules, water permeability, lateral pressure profiles, and transmembrane potential. Both low and high CHOL concentrations significantly alter lipid bilayer properties. Inducing cross-leaflet stress by mismatching lipid numbers can modify lipid order parameters and the tilting angles but has only mild effect on lateral pressure profiles and membrane thickness. Additionally, we found that transmembrane potential, generated by ions concentration differences across the membrane, can influence water permeability. Our findings expand the current understanding of lipid membrane properties and underscore the importance of considering CHOL and phospholipid asymmetry in membrane biophysics. The membrane models developed in our study also provide more physiological conditions for studying membrane proteins using MD simulations.
Human heart single-cell chromatin accessibility data reveal the diversity and complexity of heart cells at the epigenomic level, providing a detailed perspective for understanding the molecular mechanisms of heart development, function maintenance, disease occurrence, and therapeutic response. However, the current human heart single-cell chromatin accessibility data are relatively scarce, lacking large-scale, high-quality, and integrated datasets. To facilitate research and utilization, we have established a comprehensive database of human heart single-cell chromatin accessibility data, CASHeart. This database collects sequencing fragment files from publicly available papers, processes and counts data for 212,600 human heart cells, and provides transformed gene activity scores. All data are accessible for browsing and download via the online platform. We demonstrate that the data provided by CASHeart reveal heart cell type heterogeneity more effectively than the original data, aiding in the analysis of differentially accessible chromatin regions and activated genes. Moreover, we show that the incorporation of single-cell chromatin accessibility data and transformed gene activity scores from CASHeart as reference datasets enhances the analysis of heart single-cell epigenomic and transcriptomic data, whereas the unified chromatin accessible regions provided by CASHeart can assist in the study of gene regulation and genetic variation in human cardiac cells.
Single-cell genomics give us a new perspective to understand multivariate phenotypic and genetic effects at the cellular level. Recently, technologies have started measuring different modalities of individual cells, such as transcriptomes, epigenomes, metabolomes, and spatial profiling. However, integrating the results of multimodal single-cell data to identify cell-to-cell correspondences remains a challenging task. Our viewpoint emphasizes the importance of data integration at a biologically relevant level of granularity. Furthermore, it is crucial to take into account the inherent discrepancies between different modalities in order to achieve a balance between biological discovery and noise removal. In this article, we give a systematic review for the most popular single-cell integration methods and models involving cell label transfer, data visualization, and clustering task for downstream analysis. We further evaluate more than 10 popular integration methods on paired and unpaired gold standard datasets. Moreover, we discuss the data preferences of the limitations, applications, challenges and future directions of these methods.
Cell state-specific synthetic promoters are essential tools for studying and manipulating cellular function, yet their design remains challenging, particularly for complex states such as T cell exhaustion. Here we present SPECIFIC (Synthetic Promoter Engineering for Cellular State Identification and Functional Analysis), an integrated framework that leverages chromatin accessibility profiling and machine learning to systematically identify and validate cell state-specific synthetic promoters. By comparing exhausted T cells from both mouse OT-I and human CAR-T models, we identified 56 conserved transcription factor binding motifs associated with T cell exhaustion. From these motifs, we engineered a subset of the most promising candidates into synthetic promoters driving an exhaustion-responsive gene circuit that senses and responds to T cell dysfunction. Several synthetic promoters, particularly those containing NFATc2 or MEF2C binding sites, demonstrated remarkable specificity in recognizing the exhausted state and effectively attenuated T cell dysfunction by reducing both CAR expression and exhaustion markers. This study establishes a generalizable approach for designing cell state-specific regulatory elements and provides new strategies for improving CAR-T cell therapy through programmed control of gene expression.