Biology subjects

Cao, Q.

Publications and source records attributed to Cao, Q..

At least 37 records · Page 2Linked to original sources

bioRxiv2025

Brain-derived neurotrophic factor supports pericyte and vascular homeostasis in the aging brain

Microvascular circulation in the brain is often impaired in connection with the loss of pericytes in old age. The neurotrophic factor BDNF also decreases in the aging brain. We hypothesized that BDNF regulates the homeostasis of cerebral pericytes and microvasculature. We used differently aged C57BL/6J mice, and C57BL6 mice with conditional knockout of Bdnf gene. Collagen IV-positive microvessels and PDGFR{beta}-positive pericytes in the brain were counted after immunological staining. Pericytes were also quantified by Western blot of PDGFR{beta} and CD13 in isolated cerebral microvessels. The level of BDNF and TrkB phosphorylation was determined in brain homogenates. To demonstrate the direct effect of BDNF on pericytes, TrkB and pericytes were co-stained in brain tissue, single-cell sequencing and transcriptomic analysis were used to identify and characterize Ntrk2-expressing pericytes, and TrkB was also detected in the pericyte cell line by Western blot. Cultured pericytes were further treated with recombinant BDNF in the presence and absence of an Akt inhibitor and examined for PDGFR{beta} expression. The length and branching of microvessels and pericytes decreased in conjunction with the reduction in mature BDNF and TrkB phosphorylation in aging brains. Deficiency of BDNF in neurons or astrocytes was sufficient to reduce cerebral microvessels, PDGFR{beta} and CD13 concentrations and Akt and Erk1/2 phosphorylation in isolated blood vessels. A subset of pericytes in the brain and cultured pericytes expressed TrkB. BDNF treatment increased PDGFR{beta} expression along with Akt and Erk1/2 phosphorylation in cultured cells. The effect of BDNF on PDGFR{beta} expression was abolished by treatment with Akt inhibitor. Therefore, BDNF induces the expression of PDGFR{beta} and CD13 by activating Akt signaling in pericytes, promoting the homeostasis of pericytes and microvasculature in the aging brain. Our study identified a BDNF-mediated mechanism that regulates microvascular integrity in the aged brain.

neuroscience↗

bioRxiv2025

Functional recovery by transplantation of human iPSC-derived A2B5 positive neural progenitor cell after spinal cord injury in mice

Human induced pluripotent stem cells (hiPSCs) hold great potential for patient-specific therapies. Transplantation of hiPSC-derived neural progenitor cells (NPCs) is a promising reparative strategy for spinal cord injury (SCI), but clinical translation requires efficient differentiation into desired neural lineages and purification before transplantation. Here, differentiated hiPSCs--reprogrammed from human skin fibroblasts using Sendai virus-mediated expression of OCT4, SOX2, KLF4, and C-MYC--into neural rosettes expressing SOX1 and PAX6, followed by neuronal precursors ({beta}-tubulin III/NESTIN) and glial precursors (GFAP/NESTIN). Both neuronal and glial precursors expressed the A2B5 surface antigen. A2B5+ NPCs, purified by fluorescence-activated cell sorting (FACS), proliferated in vitro with mitogens and differentiated into mature neurons and astrocytes under lineage-specific conditions. NOD-SCID mice received a T9 contusion injury followed by transplantation of A2B5+ NPCs, human fibroblasts, or control medium at 8 days post-injury. At two months, grafted NPCs showed robust survival, progressive neuronal maturation ({beta}-tubulin III [->] doublecortin [->] NeuN), and astrocytic differentiation (GFAP), particularly in spared white matter. Transplantation significantly increased spared white matter volume and improved hindlimb locomotor recovery, with no teratoma formation observed. These results demonstrate that hiPSC-derived, FACS-purified A2B5+ NPCs can survive, differentiate into neurons and astrocytes, and enhance functional recovery after SCI. This approach offers a safe and effective candidate cell source for treating SCI and potentially other neurological disorders.

neuroscience↗

bioRxiv2025

Polycystin-2 is cardioprotective against myocardial infarction by regulating the calcium-mediated ER stress response.

BackgroundPatients with autosomal polycystic kidney disease (ADPKD) have an increased risk and worsened outcomes for acute myocardial infarction (AMI), but the mechanism behind this is unknown. Polycystin 2 (PC2), the protein encoded by one of the two main genes mutated in ADPKD, is a calcium-permeant channel ubiquitously expressed; however, its role in cardiomyocytes remains poorly understood. One hallmark of AMI is ER stress, which PC2 is known to regulate adaptively; however, whether PC2 regulates ER stress in the ischemic heart is unknown. ObjectiveThis study investigates the mechanism by which PC2 regulates ER stress in myocardial ischemia. Methods and ResultsPC2 protein was increased in human ischemic heart failure samples and murine myocardial infarct samples, and it was enriched at ER-mitochondrial contact sites. Induction of myocardial infarction (MI) in cardiomyocyte-specific PC2-KO mice led to cardiac dysfunction and reduced PERK expression compared to control MI mice. ER stress induced by tunicamycin in vitro blunted PERK phosphorylation and subsequent CHOP upregulation in PC2 KO cells. Tunicamycin-induced ER stress resulted in a PC2-dependent ER calcium leak and mitochondrial calcium transients, along with increased mitochondrial function, all of which were decreased in PC2 KO cells. Moreover, PC2 KO cells after ER stress exhibited decreased mitochondrial membrane potential and increased apoptosis. Isolated WT cardiomyocytes exhibited increased diastolic calcium after acute ER stress induction and increased mitochondrial uptake, neither of which was seen in PC2 KO cells. Re-expression of full-length PC2 in vitro restored both the calcium leak and PERK phosphorylation in PC2 KO cells under ER stress, but not a pathological mutant PC2 D511V, which impairs ion channel activity. ConclusionsPC2 is upregulated during ER stress, where it localizes at ER-mitochondrial contact sites and acts as an ER calcium leak channel, thereby restoring cellular homeostasis during the adaptive phase of ER stress. PC2 provides cardioprotection during ischemic events by preventing maladaptive ER stress, which contributes to cardiac dysfunction. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=194 SRC="FIGDIR/small/668747v1_ufig1.gif" ALT="Figure 1"> View larger version (43K): org.highwire.dtl.DTLVardef@1e416d5org.highwire.dtl.DTLVardef@1971fd9org.highwire.dtl.DTLVardef@abeebcorg.highwire.dtl.DTLVardef@1c68f8d_HPS_FORMAT_FIGEXP M_FIG C_FIG

physiology↗

bioRxiv2025

Towards universal modeling of transcript isoform expression levels

A holy grail in computational biology is accurate modeling of transcript expression levels using epigenetic features, which would provide a quantitative way to study gene regulation in normal and disease states. Previous studies relied heavily on immortalized cell lines that exhibit properties different from cells in natural tissue environments. Most studies also quantified the expression of each gene by a single expression level, which fails to capture separate expression levels of different transcript isoforms of the same gene. In this study, making use of the latest large-scale dataset of paired transcriptomic and epigenomic data of human samples produced by the International Human Epigenome Consortium (IHEC), we computationally modeled the expression levels of individual transcript isoforms in 324 samples from 29 tissue types. We constructed the models using graph-based methods that integrate both location-specific epigenomic features and multiple types of gene-gene relationships. We found that to infer transcript isoform expression levels in a sample, a model that integrates information from many samples of other tissue types consistently outperforms a model trained on data from this sample itself, providing strong support that it is possible to construct a "universal" model that can accurately infer transcript isoform expression levels across tissue types.

bioinformatics↗

bioRxiv2025

Copper Sulfide Nanoparticle-Mediated Photothermolysis Induces Immunogenic Cell Death in Ovarian Cancer Cells

Immunotherapy has only limited efficacy against ovarian cancer because of dysfunctional T cells in an immunosuppressive tumor microenvironment. Selective photothermolysis is a technique in which short pulsed laser is used to ablate tissue in targeted regions in a precise spatiotemporally controlled manner. This technique allows targeted tissue ablation without damage to surrounding tissue. In the current study, we demonstrated that ovarian cancer cells treated with photothermolysis mediated by near-infrared-absorbing CuS nanoparticles and 15-ns laser pulses were ingested by human monocyte-derived dendritic cells (MoDCs) more efficiently than were ovarian cancer cells treated with ionizing radiation. Moreover, ingestion of ovarian cancer cells treated with CuS NP-mediated photothermolysis promoted MoDC activation. Immature MoDCs that had been exposed to photothermolysis-treated cancer cells caused greater clonal expansion of co-cultured human CD4+ and CD8+ T cells than did immature MoDCs that had been exposed to ionizing radiation-treated tumor cells. These data indicate that photothermolysis may be used to selectively target and destroy tumor cells and cause them to release antigens and danger signals, which can be recognized by DCs to activate T cells.

immunology↗

bioRxiv2025

Cell type-specific epigenomic variation and its association with genotype in the human breast

BackgroundUnderstanding the interplay between genomic variation and the epigenome is fundamental to the study of development and mechanisms of disease. Previous studies have leveraged population-scale genotype surveys to associate alleles with epigenomic states in heterogenous tissue types. However, epigenomes are inherently cell type-specific, giving rise to unique genome-epigenome interactions that can influence distinct functional states and susceptibility to disease. Moreover, the extent of individual variation in cell type-specific epigenotypes remains poorly understood, posing additional challenges to accurately link genotypes with epigenomic features. ResultsWe generated comprehensive genomic and epigenomic profiles of four functionally defined human breast epithelial cell types from eight healthy individuals. To quantify inter-individual epigenomic variation, we developed a statistical framework that measures variability in histone modification landscapes across individuals. This analysis revealed substantially greater variation in repressive chromatin marked by H3K27me3 than in active chromatin marked by H3K27ac and H3K4me3. Integrative chromatin state analysis further identified enhancer elements as the principal source of epigenomic divergence between individuals. Stable enhancer states corresponded to high-confidence cis-regulatory elements that underpin cell type-specific transcriptional programs, whereas variable enhancer states were enriched for environmentally responsive regulatory circuits. Mapping genetic variants associated with chromatin state variation uncovered extensive cell type-specificity, with nearly 90% of regulatory variants detected in only a single cell type. These associations were strongly enriched within active regulatory chromatin and, when integrated with gene expression, enabled the prioritization of functional regulatory variants. We experimentally validated one such variant, rs75071948, demonstrating allele-specific regulation of ANXA1 expression using CRISPR/Cas9 genome editing. ConclusionsOur study defines the landscape of normal epigenomic variation across the major human breast epithelial cell types and demonstrates that genome-epigenome interactions are highly cell type-specific. These findings establish cell type as a critical determinant of the functional interpretation of regulatory genetic variation and provide a framework for understanding how inherited genetic variation shapes normal breast biology and disease susceptibility.

genomics↗

bioRxiv2025

Design and optimization of a kinase-controlled allosteric switch

Post-translational control enables rapid and precise regulation of cell behavior. Despite these advantages, general strategies to build phosphorylation-based synthetic circuits are limited. Here, we reasoned that engineered allostery, a technique that has been applied to design light- and chemically-gated protein switches, could also be used to engineer phosphorylation-controlled protein switches (phospho-switches). Using an allosterically controllable Gal4 transcription factor as a scaffold, we show that a classic kinase FRET biosensor architecture can be used as a starting point for phospho-switch design. We optimize all features of the phospho-switch to develop an ERK-controlled transcription factor with a 20-fold phosphorylation-dependent change in transcriptional output. Our switch outperforms the c-fos promoter, a classic ERK-responsive transcriptional biosensor, in selectivity and sensitivity. We further show that our switch architecture can be generalized to other input kinases and allosterically controlled targets. This work provides a general platform for a new generation of kinase-responsive tools for biosensing and synthetic biology applications.

synthetic biology↗

bioRxiv2025

Few-shot in-context learning with large language models for antibody characterization

Large language models (LLMs) can learn new tasks by in-context learning (ICL), but it is unknown whether this ability reliably transfers to biological sequence classification. Here, we systematically evaluate how demonstration selection, shot count, and prompting strategies affect performance across 20 general-purpose LLMs. Using antibody characterization as a representative test case, we compare zero-shot, few-shot, and chain-of-thought (CoT) ICL on three classification tasks: humanness, antigen specificity, and isotype. Our results reveal a clear performance hierarchy: while zero-shot prompting performs near chance, few-shot prompting with randomly selected demonstrations improves performance, showing that LLMs can perform ICL using biological sequences from minimal supervision. However, matching protein-language model (pLM)-based classifier accuracy is only achieved when using label-diverse demonstrations drawn from antibodies similar to the query sequence. To leverage this insight, we introduce Sim-ICL, a framework that automatically retrieves such demonstrations. Using only 32-shot prompting, Sim-ICL achieves performance competitive with pLM-based classifiers, matching or outperforming them in two of the three tasks. Furthermore, reasoning-oriented prompts yield marginal gains and often produce fluent but biologically incorrect rationales, suggesting that current CoT explanations function as after-the-fact rationalizations rather than capturing mechanistic determinants of antibody properties. From these experiments, we derive practical design principles for ICL on biological sequences: use similarity-based, label-diverse demonstrations and modest shot counts, and treat reasoning prompts primarily as post hoc narratives rather than drivers of performance. Sim-ICL implements these principles in a streamlined, prompt-based framework for antibody sequence classification and, in principle, could be adapted to other biological sequence tasks.

bioinformatics↗

bioRxiv2025

WormAI: Artificial Intelligence of Networks for Nematode Phenotyping

The nematode Caenorhabditis elegans is a key model organism in biological research due to its genetic similarity to humans and its utility in studying complex processes. Traditional image analysis methods, such as those using ImageJ, are labour-intensive, which has led to the integration of AI. This study introduces an AI framework with three machine learning models: WormGAN, a Generative Adversarial Network for generating synthetic nematode images to enhance training data; WormDET, for precise movement tracking; and WormREG, for accurate anatomical measurements. Together, these tools significantly improve the efficiency and accuracy of phenotypic analysis. WormAI demonstrates substantial potential for high-throughput dataset analysis, advancing research in systems biology, drug discovery, and aging. This framework streamlines workflows, enabling faster and more precise discoveries in C. elegans studies.

bioinformatics↗

bioRxiv2025

Dnmt3b deficiency in adipocyte progenitor cells ameliorates obesity in female mice

Obesity arises from chronic energy imbalance where energy intake exceeds energy expenditure. Emerging evidence supports a key role of DNA methylation in the regulation of adipose tissue development and metabolism. We recently discovered a key role of DNA methylation, catalyzed by DNA methyltransferase 1 or 3a (Dnmt1 or 3a), in the regulation of adipocyte differentiation and metabolism. Here, we aimed to investigate the role of adipocyte progenitor cell Dnmt3b, an enzyme mediating de novo DNA methylation, in energy metabolism and obesity. We generated a genetic model with Dnmt3b knockout in adipocyte progenitor cells (PD3bKO) by crossing Dnmt3b-floxed mice with platelet-derived growth factor receptor alpha (Pdgfr)-Cre mice. Dnmt3b gene deletion in adipocyte progenitors enhanced thermogenic gene expression in brown adipose tissue, increased overall energy expenditure, and mitigated high-fat diet (HFD)-induced obesity in female mice. PD3bKO mice also displayed a lower respiratory exchange ratio (RER), indicative of a metabolic shift favoring fat utilization as an energy source. Furthermore, female PD3bKO mice exhibited improved insulin sensitivity alongside their lean phenotype. In contrast, male PD3bKO mice showed no changes in body weight but demonstrated decreased insulin sensitivity, revealing a sexually dimorphic metabolic response to Dnmt3b deletion in adipocyte progenitor cells. These findings underscore the critical role of Dnmt3b in regulating energy homeostasis, body weight, and metabolic health, with significant implications for understanding sex-specific mechanisms of obesity and metabolism.

systems biology↗

bioRxiv2025

Painting Peptides with Antimicrobial Potency through Deep Reinforcement Learning

In the post-antibiotic era, antimicrobial peptides (AMPs) serve as ideal drug candidates for their lower likelihood of inducing resistance. Computational models offer an efficient way to design novel AMPs. However, current optimization and generation approaches are tailored for different application scenarios. To address this challenge, we propose a novel AMP design model named AMPainter. Based on deep reinforcement learning, AMPainter integrates both optimization and generation tasks in a unified framework. We apply AM-Painter to three types of peptides, including known AMPs, signal peptides (SPs), and random sequences. AMPainter outperforms ten related models in enhancing the activity of known AMPs, and evolves effective AMPs from membrane-active SPs with a success rate of 80%. Furthermore, several de novo designed AMPs from random sequences are validated along with their evolutionary paths. Therefore, AMPainter contributes to paint antimicrobial potency to diverse peptides, assisting in expanding the AMP sequence space and discovering novel antimicrobial agents.

bioinformatics↗

bioRxiv2024

Effect of Stroma-directed Drugs in Combination with Chemotherapy Against Pancreatic Cancer- a Preclinical Study

Cytotoxic chemotherapy plays an important role for extending the survival of patients with pancreatic ductal adenocarcinoma (PDAC). To enhance the efficacy of chemotherapy for eradicating the cancer cells, we have compared the standard care chemotherapy (combination of nab-paclitaxel, gemcitabine and cisplatin, NGC) versus NGC plus stroma-directed agents (calcipotriol and losartan, respectively) in a genetically engineered mouse model of PDAC. Over a 2-week study period, MRI was conducted to measure the tumor size and to test the sensitivity of imaging markers derived from diffusion-weighted imaging (DWI), dynamic contrast enhanced MRI (DCE) and magnetization transfer ratio (MTR) for assessing the tumor cellularity and stromal changes. Detailed immunohistochemistry and preliminary single cell RNA sequencing (scRNAseq) study were applied to tumor tissues collected upon euthanasia on day-14. Our major findings are: 1. Compared the untreated controls, NGC chemotherapy induced significant tumor growth inhibition and stromal changes including pronounced reduction of fibroblast associated protein (FAP) level accompanied by increased matrix collagen content, significantly reduced microvascular permeability revealed by DCE corroborated with reduced microvascular density. 2. Losartan+NGC significantly enhanced inhibition of tumor growth beyond NGC and increased lymphocytes infiltration in the tumor which may contribute to enhanced cancer cells eradication. 3. NGC treatment enriched the fraction of mesenchymal (M) subtype while reducing the epithelial (E) subtype of cancer cells compared to the controls, and this trend was reversed by calcipotriol+NGC. In conclusion, our study captured changes in cancer cell and tumor microenvironment in response to chemo stromal therapy versus chemotherapy alone with mechanistic insights.

cancer biology↗

bioRxiv2024

Zero valent sulfur is the product of assimilatory sulfite reductase and a substrate of cysteine synthase

The long process of the assimilatory sulfate-reduction consists of sulfate reduction to H2S and then H2S assimilation for cysteine synthesis. However, recent studies indicate that both H2S and zero valent sulfur (S0) are produced during dissimilatory sulfate reduction process, which leads to the question whether S0 also is involved in assimilatory sulfate-reduction. Herein, phylogenetic analysis grouped assimilatory sulfite reductases into five clusters and we re-examined four representatives from two major clusters. These enzymes produced both S0 and H2S from sulfite. However, S0 was more like a direct product, while H2S was just a derivative of S0 via chemical reduction. Phylogenetic analysis of cysteine synthases grouped them into six clusters, and we re-examined six representatives from two major clusters. These enzymes used both glutathione persulfide (GSSH, a common intracellular compound containing a S0 atom) and H2S as substrates to synthesize cysteine, but preferred to use GSSH. These findings indicated that S0, other than H2S, is the key intermediate during the process of inorganic sulfur converting to organic sulfur. ImportanceSulfate reduction is an important link in the global sulfur cycle. Assimilatory sulfate-reduction consists of sulfate-reduction to H2S and then H2S assimilation for cysteine synthesis. This concept was founded in 1950s and has not been challenged until today. In this study, we re-examined four representative assimilatory sulfite reductases (aSiRs) and found that these enzymes directly reduced sulfite to zero valent sulfur (S0). A small portion of produced S0 was reduced to H2S via aSiR-independent chemical reactions. Further, we re-examined six representative cysteine synthases and found that these enzymes used S0 derivative compounds (thiosulfate and glutathione persulfide) as substrates to synthesize cysteine. Thus, S0 is the key intermediate of assimilatory sulfate reduction process, other than H2S.

microbiology↗

bioRxiv2024

Quantitative MRI Measurements Capture Pancreatic Cancer and Stroma Reactions to New KRAS Inhibitor

PurposeIn pancreatic ductal adenocarcinoma (PDAC), KRAS mutations drive both cancer cell growth and formation of a dense stroma. Small molecule KRAS inhibitors (KRASi) represent a promising new treatment hence clinical tools that can assess early response, detect resistance and/or predict prolonged survival are desirable to understand clinical biology of KRASi. We hypothesized that diffusion-weighted MRI (DWI) can detect cell death while dynamic contrast enhanced MRI (DCE) and magnetization transfer ratio (MTR) imaging are sensitive to tumor microenvironment changes, and these metrics shed insights into tumor size change induced by KRASi treatment. Experimental DesignMultiple preclinical PDAC models including a genetically engineered mouse model (KPC) received MRTX1133, a KRASi specific for KRASG12D mutation. Quantitative imaging markers were corroborated with immunohistochemistry (IHC) analyses. ResultsSignificant increase of tumor apparent diffusion coefficient (a DWI metric) was detected as early as 48h and persisted to Day7 after initiation of KRASi treatment and was strongly correlated with cell death and reduced cellularity, resulting in greatly prolonged median survival in treated mice. Capillary perfusion/permeability (a DCE metric) exhibited an inverse relationship with microvascular density. Distinct responses of KRASG12C versus KRASG12D tumors to MRTX1133 were captured by the MRI metrics corroborated with IHC. When tumors developed resistance to MRTX1133, the imaging marker values exhibited a reversal from those of responding tumors. ConclusionsMultiparametric MRI provides early biological insights of cancer and stromal response to KRASi treatment and sets the stage for testing the utility of these clinically ready MRI methods in patients receiving KRASi therapy. Translational relevanceEmerging small molecule KRAS inhibitors (KRASi) represent a new class of therapy for PDAC. Clinical tools that can provide early biological insights of KRASi therapy are desirable. In PDAC models, we examined a clinically ready imaging protocol that combines MRI-based tumor size, diffusion-weighted MRI (DWI), dynamic contrast enhanced MRI (DCE), and magnetization transfer ratio (MTR) for detection of early response as well as acquired resistance to MRTX1133, a KRASi being evaluated in clinical trials. Our data show that DWI and DCE metrics provided key insights of significant cell death and tumor microenvironment changes underlying tumor size regression as early as 48 hours after KRASi treatment initiation. These MRI metrics also captured resistance to KRASi developed over prolonged treatment. This study has high translational relevance by employing clinically applied MRI methods, an investigational new drug and a genetically engineered mouse model that recapitulates salient features of human PDAC.

cancer biology↗

bioRxiv2024

A motif preferred adenine base editor with minimal bystander and off-targets editing

48% of hereditary disease are caused by single C-to-T base conversion, which makes efficient A-to-G base editing tools (ABEs) have great potential in the treatment of these diseases. However, the existing efficient ABE, while catalyzing A-to-G conversion, will bring more A and C bystander editing and off-target events, which poses safety concerns for their clinical application. Here, we developed ABE10 (ABE8e with TadA-8e A48E) for efficient and accurate editing of As in YA motifs with YAY>YAR (Y=T or C, R=A or G) hierarchy through structure-oriented rational design. Compared with ABE3.1, currently the only motif (YAC) preference ABE version, ABE10 exhibited A-to-G editing efficiency improvement with an average up to 3.1-fold in indicated YA motif while maintaining reduced bystander Cs editing and minimized DNA or RNA off-targets. Also, we showed ABE10 corrected pathogenic mutation with high efficiency and precision in human cells. Moreover, by ABE10, we efficiently and precisely generated hypocholes-terolemia and tail-loss mouse models mimicking human associated disease and mouse PCSK9 base editing in vivo for hypercholesterolemia gene therapy, indicating their great potential in broad applications for and disease modeling and gene therapy.

molecular biology↗

bioRxiv2024

A Genomic Language Model for Chimera Artifact Detection in Nanopore Direct RNA Sequencing

Chimera artifacts in nanopore direct RNA sequencing (dRNA-seq) can significantly distort transcriptome analyses, yet their detection and removal remain challenging due to limitations in existing basecalling models. We present Deep-Chopper, a genomic language model that precisely identifies and removes adapter sequences from base-called dRNA-seq long reads at single-base resolution, operating independently of raw signal or alignment information to effectively eliminate chimeric read artifacts. By removing these artifacts, DeepChopper substantially improves the accuracy of critical downstream analyses, such as transcript annotation and gene fusion detection, thereby enhancing the reliability and utility of nanopore dRNA-seq for transcriptomics research.

genomics↗

bioRxiv2024

Overcoming artificial structures in resolution-enhanced Hi-C data by signal decomposition and multi-scale attention

Computational enhancement is an important strategy for inferring high-resolution features from genome-wide chromosome conformation capture (Hi-C) data, which typically have limited resolution. Deep learning has been highly successful in this task but we show that it creates prevalent artificial structures in the enhanced data due to the need to divide the large contact matrix into small patches. In addition, previous deep learning methods largely focus on local patterns, which cannot fully capture the complexity of Hi-C data. Here we propose Smooth, High-resolution, and Accurate Reconstruction of Patterns (SHARP) for enhancing Hi-C data. It uses the novel approach of decomposing the data into three types of signals, due to one-dimensional proximity, contiguous domains, and other fine structures, and applies deep learning only to the third type of signals, such that enhancement of the first two is unaffected by the patches. For the deep learning part, SHARP uses both local and global attention mechanisms to capture multi-scale contextual information. We compare SHARP with state-of-the-art methods extensively, including application to data from new samples and another species, and show that SHARP has superior performance in terms of resolution enhancement accuracy, avoiding creation of artificial structures, identifying significant interactions, and enrichment in chromatin states.

bioinformatics↗

bioRxiv2024

Subcellular Level Spatial Transcriptomics with PHOTON

The subcellular localization of RNA is closely linked to its function. Many RNA species are partitioned into organelles and other subcellular compartments for storage, processing, translation, or degradation. Thus, capturing the subcellular spatial distribution of RNA would directly contribute to the understanding of RNA functions and regulation. Here, we present PHOTON (Photoselection of Transcriptome over Nanoscale), a method which combines high resolution imaging with high throughput sequencing to achieve spatial transcriptome profiling at subcellular resolution. We demonstrate PHOTON as a versatile tool to accurately capture the transcriptome of target cell types in situ at the tissue level such as granulosa cells in the ovary, as well as RNA content within subcellular compartments such as the nucleolus and the stress granule. Using PHOTON, we also reveal the functional role of m6A modification on mRNA partitioning into stress granules. These results collectively demonstrate that PHOTON is a flexible and generalizable platform for understanding subcellular molecular dynamics through the transcriptomic lens.

genomics↗