Dictionary learning allows model-free pseudotime estimation of transcriptomic data

Background Pseudotime estimation from dynamic single-cell transcriptomic data enables characterisation and understanding of the underlying processes, for example developmental processes. Various pseudotime estimation methods have been proposed during the last years. Typically, these methods start with a dimension reduction step because the low-dimensional representation is usually easier to analyse. Approaches such as PCA, ICA or t-SNE belong to the most widely used methods for dimension reduction in pseudotime estimation methods. However, these methods usually make assumptions on the derived dimensions, which can result in important dataset properties being missed. In this paper, we suggest a new dictionary learning based approach, dynDLT, for dimension reduction and pseudotime estimation of dynamic transcriptomic data. Dictionary learning is a matrix factorisation approach that does not restrict the dependence of the derived dimensions. To evaluate the performance, we conduct a large simulation study and analyse 8 real-world datasets. Results The simulation studies reveal that firstly, dynDLT preserves the simulated patterns in low-dimension and the pseudotimes can be derived from the low-dimensional representation. Secondly, the results show that dynDLT is suitable for the detection of genes exhibiting the simulated dynamic patterns, thereby facilitating the interpretation of the compressed representation and thus the dynamic processes. For the real-world data analysis, we select datasets with samples that are taken at different time points throughout an experiment. The pseudotimes found by dynDLT have high correlations with the experimental times. We compare the results to other approaches used in pseudotime estimation, or those that are method-wise closely connected to dictionary learning: ICA, NMF, PCA, t-SNE, and UMAP. DynDLT has the best overall performance for the simulated and real-world datasets. Conclusions We introduce dynDLT, a method that is suitable for pseudotime estimation. Its main advantages are: (1) It presents a model-free approach, meaning that it does not restrict the dependence of the derived dimensions; (2) Genes that are relevant in the detected dynamic processes can be identified from the dictionary matrix; (3) By a restriction of the dictionary entries to positive values, the dictionary atoms are highly interpretable.

Construction of the Classification Model Using Key Genes Identified Between Benign and Malignant Thyroid Nodules From Comprehensive Transcriptomic Data

Thyroid nodules are present in upto 50% of the population worldwide, and thyroid malignancy occurs in only 5–15% of nodules. Until now, fine-needle biopsy with cytologic evaluation remains the diagnostic choice to determine the risk of malignancy, yet it fails to discriminate as benign or malignant in one-third of cases. In order to improve the diagnostic accuracy and reliability, molecular testing based on transcriptomic data has developed rapidly. However, gene signatures of thyroid nodules identified in a plenty of transcriptomic studies are highly inconsistent and extremely difficult to be applied in clinical application. Therefore, it is highly necessary to identify consistent signatures to discriminate benign or malignant thyroid nodules. In this study, five independent transcriptomic studies were combined to discover the gene signature between benign and malignant thyroid nodules. This combined dataset comprises 150 malignant and 93 benign thyroid samples. Then, there were 279 differentially expressed genes (DEGs) discovered by the feature selection method (Student’s t test and fold change). And the weighted gene co-expression network analysis (WGCNA) was performed to identify the modules of highly co-expressed genes, and 454 genes in the gray module were discovered as the hub genes. The intersection between DEGs by the feature selection method and hub genes in the WGCNA model was identified as the key genes for thyroid nodules. Finally, four key genes (ST3GAL5, NRCAM, MT1F, and PROS1) participated in the pathogenesis of malignant thyroid nodules were validated using an independent dataset. Moreover, a high-performance classification model for discriminating thyroid nodules was constructed using these key genes. All in all, this study might provide a new insight into the key differentiation of benign and malignant thyroid nodules.

Loss of plastid developmental genes coincides with a reversion to monoplastidy in hornworts

The first plastid evolved from an endosymbiotic cyanobacterium in the common ancestor of the Archaeplastida. The transformative steps from cyanobacterium to organelle included the transfer of control over developmental processes; a necessity for the host to orchestrate, for example, the fission of the organelle. The plastids of almost all embryophytes divide independent from nuclear division, leading to cells housing multiple plastids. Hornworts, however, are monoplastidic (or near-monoplastidic) and their photosynthetic organelles are a curious exception among embryophytes for reasons such as the occasional presence of pyrenoids. Here we screened genomic and transcriptomic data of eleven hornworts for components of plastid developmental pathways. We find intriguing differences among hornworts and specifically highlight that pathway components involved in regulating plastid development and biogenesis were differentially lost in this group of bryophytes. In combination with ancestral state reconstruction, our data suggest that hornworts have reverted back to a monoplastidic phenotype due to the combined loss of two plastid division-associated genes: ARC3 and FtsZ2.

Cell type-specific aging clocks to quantify aging and rejuvenation in regenerative regions of the brain

Aging manifests as progressive dysfunction culminating in death. The diversity of cell types is a challenge to the precise quantification of aging and its reversal. Here we develop a suite of 'aging clocks' based on single cell transcriptomic data to characterize cell type-specific aging and rejuvenation strategies. The subventricular zone (SVZ) neurogenic region contains many cell types and provides an excellent system to study cell-level tissue aging and regeneration. We generated 21,458 single-cell transcriptomes from the neurogenic regions of 28 mice, tiling ages from young to old. With these data, we trained a suite of single cell-based regression models (aging clocks) to predict both chronological age (passage of time) and biological age (fitness, in this case the proliferative capacity of the neurogenic region). Both types of clocks perform well on independent cohorts of mice. Genes underlying the single cell-based aging clocks are mostly cell-type specific, but also include a few shared genes in the interferon and lipid metabolism pathways. We used these single cell-based aging clocks to measure transcriptomic rejuvenation, by generating single cell RNA-seq datasets of SVZ neurogenic regions for two interventions - heterochronic parabiosis (young blood) and exercise. Interestingly, the use of aging clocks reveals that both heterochronic parabiosis and exercise reverse transcriptomic aging in the niche, but in different ways across cell types and genes. This study represents the first development of high-resolution aging clocks from single cell transcriptomic data and demonstrates their application to quantify transcriptomic rejuvenation.

Integration of Aspergillus niger transcriptomic profile with metabolic model identifies potential targets to optimise citric acid production from lignocellulosic hydrolysate

Background Citric acid is typically produced industrially by Aspergillus niger-mediated fermentation of a sucrose-based feedstock, such as molasses. The fungus Aspergillus niger has the potential to utilise lignocellulosic biomass, such as bagasse, for industrial-scale citric acid production, but realising this potential requires strain optimisation. Systems biology can accelerate strain engineering by systematic target identification, facilitated by methods for the integration of omics data into a high-quality metabolic model. In this work, we perform transcriptomic analysis to determine the temporal expression changes during fermentation of bagasse hydrolysate and develop an evolutionary algorithm to integrate the transcriptomic data with the available metabolic model to identify potential targets for strain engineering. Results The novel integrated procedure matures our understanding of suboptimal citric acid production and reveals potential targets for strain engineering, including targets consistent with the literature such as the up-regulation of citrate export and pyruvate carboxylase as well as novel targets such as the down-regulation of inorganic diphosphatase. Conclusions In this study, we demonstrate the production of citric acid from lignocellulosic hydrolysate and show how transcriptomic data across multiple timepoints can be coupled with evolutionary and metabolic modelling to identify potential targets for further engineering to maximise productivity from a chosen feedstock. The in silico strategies employed in this study can be applied to other biotechnological goals, assisting efforts to harness the potential of microorganisms for bio-based production of valuable chemicals.

LncRNAs Are Differentially Expressed between Wildtype and Cell Line Strains of African Trypanosomes

Trypanosoma brucei is a parasitic protist that causes African sleeping sickness. The establishment of T. brucei cell lines has provided a significant advantage for the majority of T. brucei research. However, these cell lines were isolated and maintained in culture for decades, occasionally accumulating changes in gene expression. Since trypanosome strains have been maintained in culture for decades, it is possible that difference may have accumulated in fast-evolving non-coding RNAs between trypanosomes from the wild and those maintained extensively in cultures. To address this, we compared the lncRNA expression profile of trypanosomes maintained as cultured cell lines (CL) to those extracted from human patients, wildtype (WT). We identified lncRNAs from CL and WT from available transcriptomic data and demonstrate that CL and WT have unique sets of lncRNAs expressed. We further demonstrate that the unique and shared lncRNAs are differentially expressed between CL and WT parasites, and that these lncRNAs are more evenly up-regulated and down-regulated than protein-coding genes. We validated the expression of these lncRNAs using qPCR. Taken together, this study demonstrates that lncRNAs are differentially expressed between cell lines and wildtype T. brucei and provides evidence for potential evolution of lncRNAs, specifically in T. brucei maintained in culture.

STEEL enables high-resolution delineation of spatiotemporal transcriptomic data

Advances in spatial transcriptomics enlarge the use of single cell technologies to unveil the expression landscape of the tissues with valuable spatial context. However, computational tools developed for single-cell transcriptomics have great limits in dealing with spatial transcriptomic data with high noise on detected transcript signals. Here we propose an unsupervised and manifold learning-based algorithm, STEEL, which identifies different cell types from spatial transcriptome by clustering cells/beads exhibiting both highly similar gene expression profiles and close spatial distance in the manner of graphs. Comprehensive evaluation of STEEL on various spatial transcriptomic datasets from 10X Visium platform demonstrates that it not only achieves a high resolution to characterize fine structures of mouse brain, but also enables the integration of multiple tissue slides individually analyzed into a larger one. STEEL outperforms previous methods to effectively distinguish different cell types of various tissues on Slide-seq datasets, featuring in higher bead density but lower transcript detection efficiency. Application of STEEL on spatial transcriptomes of early-stage mouse embryos (E9.5 to E12.5) successfully delineates a progressive development landscape of tissues from ectoderm, mesoderm and endoderm layers, and futher profiles dynamic changes on cell differentiation in heart and other organs. With the advancement of spatial transcriptome technologies, our method will have great applicability in high-resolution cell type identification and unbiased spatiotemporal data integration.

Analysis of the glyco-code in pancreatic ductal adenocarcinoma identifies glycan-mediated immune regulatory circuits

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most aggressive malignancies with a 5-year survival rate of only 9%. Despite the fact that changes in glycosylation patterns during tumour progression have been reported, no systematic approach has been conducted to evaluate its potential for patient stratification. By analysing publicly available transcriptomic data of patient samples and cell lines, we identified here two specific glycan profiles in PDAC that correlated with progression, clinical outcome and epithelial to mesenchymal transition (EMT) status. These different glycan profiles, confirmed by glycomics, can be distinguished by the expression of O-glycan fucosylated structures, present only in epithelial cells and regulated by the expression of GALNT3. Moreover, these fucosylated glycans can serve as ligands for DC-SIGN positive tumour-associated macrophages, modulating their activation and inducing the production of IL-10. Our results show mechanisms by which the glyco-code contributes to the tolerogenic microenvironment in PDAC.

EchinoDB: An update to the web-based application for genomic and transcriptomic data on Echinoderms.

Here we release a new version of EchinoDB (https://echinodb.uncc.edu). EchinoDB is a database of genomic and transcriptomic data on echinoderms. The initial database consisted of groups of 749,397 orthologous and paralogous transcripts arranged in orthoclusters by sequence similarity. The new version of EchinoDB includes RNA-seq data of the brittle star Ophioderma brevispinum and high-quality genomic assembly data of the green sea urchin Lytechinus variegatus. In addition, we enabled keyword searches for annotated data and installed an updated version of Sequenceserver to allow BLAST searches. The data are downloadable in FASTA format. The first version of EchinoDB appeared in 2016 and was implemented in GO on a local server. The new version has been updated using R Shiny to include new features and improvements in the application. Furthermore, EchinoDB now runs entirely in the cloud for increased reliability and scaling. EchinoDB enjoys a user base drawn from the fields of phylogenetics, developmental biology, genomics, physiology, neurobiology, and regeneration. As use cases, we illustrate how EchinoDB is used in discovering pathways and gene regulatory networks involved in the tissue regeneration process.