bioRxiv Science⌕ Search

Biology subjects

Hewitt, A. W.

Publications and source records attributed to Hewitt, A. W..

At least 19 recordsLinked to original sources

Computational drug repurposing identifies N-acetylglucosamine as a potential therapeutic compound for CLN3 Batten disease

Batten disease, also known as neuronal ceroid lipofuscinoses, is one of the most common causes of childhood dementia. It is characterized by the accumulation of lipofuscin in lysosomes, leading to loss of brain cell function, onset of dementia-like symptoms, vision loss and seizures and has extremely limited treatment options. Here, we performed computational drug repurposing analysis to identify existing compounds that may target Batten disease risk genes. A total of 81 candidate compounds were identified, 6 of which were selected based on clinical tractability for downstream testing in Batten disease (CLN3) iPSC-derived models. After confirming disease phenotype and drug candidate safety, CLN3 brain cell cultures treated with and without drug candidates underwent bulk RNA-seq to identify drug responses. One of the candidate drugs N-acetylglucosamine (GlcNAc) significantly upregulated Batten disease risk gene CLN5 expression and several other lysosomal markers within CLN3 brain cells, and modulated several pathways implicated in lysosomal storage disorders. Importantly, GlcNAc significantly reduced lipofuscin burden in both CLN3 iPSC-derived neurons and astrocytes, supporting its investigation as an additional therapy for Batten disease.

neuroscience↗

Optimization of AAV tools to target M&uumlller glial cells for retinal gene therapy

Reprogramming of Muller glial (MG) cells into retinal neurons has the potential to treat vision loss by regenerating the retina. Development of efficient gene delivery systems to target the MG cells is critical. Adeno-associated virus (AAV) serotypes and promoter specificity are important factors that influence AAV transduction profile in the retina. However, studies that optimize these parameters to specifically target MG cells are limited, in particular in rats which are commonly used for eye research. Here we tested 4 AAV serotypes and 14 promoters to optimize gene delivery to human MG cells in vitro and/or rat MG cells in vivo. We showed that the combinatorial use of MG-specific serotypes and promoters achieved high specificity for MG cell targeting, with ShH10Y serotype and the GFAP (gfaABC1D) promoter as the best performing tool to target rat MG cells in vivo. We developed new AAV vectors using known and novel MG-specific promoters and engineered short promoter variants to improve the cargo capacity of AAV delivery. Our results highlighted a number of promoters that can target MG cells in vitro or in vivo. This study further expands the AAV toolbox to target MG cells, which has important implications for retinal gene therapy development.

bioengineering↗

Optimization of AAV tools to target Muller glial cells for retinal gene therapy

Reprogramming of Muller glial (MG) cells into retinal neurons has the potential to treat vision loss by regenerating the retina. Development of efficient gene delivery systems to target the MG cells is critical. Adeno-associated virus (AAV) serotypes and promoter specificity are important factors that influence AAV transduction profile in the retina. However, studies that optimize these parameters to specifically target MG cells are limited, in particular in rats which are commonly used for eye research. Here we tested 4 AAV serotypes and 14 promoters to optimize gene delivery to human MG cells in vitro and/or rat MG cells in vivo. We showed that the combinatorial use of MG-specific serotypes and promoters achieved high specificity for MG cell targeting, with ShH10Y serotype and the GFAP (gfaABC1D) promoter as the best performing tool to target rat MG cells in vivo. We developed new AAV vectors using known and novel MG-specific promoters and engineered short promoter variants to improve the cargo capacity of AAV delivery. Our results highlighted a number of promoters that can target MG cells in vitro or in vivo. This study further expands the AAV toolbox to target MG cells, which has important implications for retinal gene therapy development.

molecular biology↗

Patient induced pluripotent stem cells identify specificities of a reticular pseudodrusen phenotype in age-related macular degeneration

BackgroundAge-related macular degeneration (AMD) is a leading cause of vision loss. Reticular pseudodrusen (RPD), deposits on the apical side of the retinal pigment epithelium (RPE), signify a distinctive and critical AMD phenotype. Yet, their molecular basis and relationship to the conventional drusen seen in AMD remain unclear. ResultsWe generated induced pluripotent stem cell-derived RPE cells from a clinically phenotyped cohort comprising only individuals with conventional drusen (AMD/RPD-) or drusen coexisting with RPD (AMD/RPD+). From these cells, we generated single-cell transcriptomics, proteomics, and functional data to identify differences between the two cohorts. We show that AMD/RPD+ RPE cells exhibit enrichment in extracellular matrix (ECM) remodelling, cytoskeletal, and hypoxia-responsive programs, whereas AMD/RPD- RPE cells display a relatively greater representation of mitochondrial and protein homeostasis pathways. Both subtypes engaged pathways classically linked to ageing, including ECM remodelling and mitochondrial function, but differed in the direction and extent of these changes. Expression and protein quantitative trait loci (QTLs) highlight shared genetic influences on mitochondrial and iron-handling pathways, while disease-interacting eQTLs and transcriptome-wide association study identify regulatory signals that are distinctive of the RPD subtype within AMD, including through regulation of ECM. Functionally, all iPSC-derived RPE formed drusen-like deposits in vitro: AMD/RPD-lines generated more basal deposits, whereas AMD/RPD+ cells exhibited greater structural instability under bisretinoid-induced stress. ConclusionsThese findings indicate that AMD with and without RPD represent mechanistically distinct entities and provide novel insight into the molecular mechanisms underlying disease heterogeneity in AMD.

genetics↗

Exploring Prime Editing strategies for genetic correction of a Ceroid Neuronal Lipofuscinosis type 2 disease causing variant

Neuronal ceroid lipofuscinosis type 2 (CLN2) is a rare, fatal paediatric neurodegenerative genetic disease that causes progressive psychomotor decline, epilepsy, speech impairment, vision loss, and premature death by late childhood or early adolescence. It is caused by mutations in the TPP1 gene, which encodes the tripeptidyl peptidase 1 enzyme. To date, enzyme replacement therapy (ERT) with recombinant human TPP1 is the only approved treatment available and it does not appear to have a curative effect on disease progression or rescuing complex motor functions. This underscores the importance of developing more effective treatment strategies for CLN2 disease. In the quest for an alternative therapy, we here seek to utilise CRISPR based prime editing (PE) to correct one of the most frequent TPP1 mutations of c.622 C>T (p. Arg208Ter). In cultured mammalian HEK293a cells we screened 16 different PE and pegRNA combinations, achieving on-target editing efficiencies up to 27.36 %. To enable efficient in vivo gene editing, we packaged the most promising PE and pegRNA combination using engineered virus-like particles (eVLPs) which resulted in 12.6 % on-target editing. Given the robust on-target editing and reduced risk of bystander editing, PE is deemed suitable for in vivo testing in the pre-clinical models for CLN2 disease. Overall, our findings establish the proof of concept for CRISPR-based prime editing to correct pathogenic human TPP1 nonsense mutation of c.622 C>T (p. R208X) and provide a basis for further investigations of PE as a genetic treatment for CLN2 disease.

molecular biology↗

Engineering single-AAV CRISPR-Cas13 RNA base editors for treatment of inherited retinal diseases

Site-directed RNA editing, especially RNA base editing, allows for specific manipulation of RNA sequences, making it a useful approach for the correction of pathogenic mutations. Correction of RNA transcripts allows therapeutic gene editing in a safe and reversible manner and avoids permanent alterations in the genome. RNA-targeting CRISPR-Cas nucleases (e.g., CRISPR-Cas13) enable delivery within a single adeno-associated virus (AAV) vector for RNA base editing, making the approach clinically feasible. Here, we used the inactive CRISPR-Cas13bt3 (also known as Cas13X.1) fused to the ADAR2 deaminase domain (ADAR2DD) for targeted correction of inherited retinal disease (IRD) mutations. First, we show in vitro that dCas13bt3-ADAR2DD can efficiently correct a pathogenic nonsense mutation (c.130C>T [p.R44X]) found in the mouse Rpe65 gene and recover protein expression in retinal pigment epithelium cells (RPEs). Across clinically reported RPE65 mutations, we observed editing efficiencies ranging from 0% to 60%. In the Rpe65-deficient mouse model of retinal degeneration (rd12), we observed that RNA base editing can recover Rpe65 expression in RPEs and rescue retinal function with no observable adverse effects. We further employed our RNA base editor against the large USH2A gene to assess the promise of RNA base editing for addressing untreatable IRDs caused by genes too large for AAV gene delivery. Against the human USH2A in vitro, we observed up to 60% on-target efficiency. We further found that gRNA mismatches, domain-inlaid ADAR2DD design and nucleocytoplasmic shuttling of the RNA base editor optimised on-target and bystander editing for a highly precise base editor. Against the mouse Ush2a in vitro, we similarly observed up to 60% on-target editing in mammalian cells, while in the Ush2aW3947X mice, we observed [~]12% on-target editing, with no impact on retinal structure or function, or transcriptome-wide editing. Overall, our findings demonstrate dCas13bt3-ADAR2DD as a potent tool for gene therapy against IRDs, addressing a significant unmet clinical need in ophthalmology.

bioengineering↗

The impact of sex on the immune system explored at the single-cell level

Sex has a key role in disease susceptibility, in particular autoimmunity. Sex differences of the immune system stems from genes and their interactions with intrinsic and external factors. However, the cellular-level factors influencing sexual dimorphism is not fully understood. We thus examined immune sex differences at single-cell resolution to dissect the genetic impacts. Female-biased sex-differentially expressed genes (DEGs) in multiple immune cells were involved in TNF[a] signalling, whereas male DEGs were enriched for ribosomal-related functions. While cis-eQTLs were less common on sex chromosomes, we identified over 1000 sex-specific eQTLs and 51 sex-interacting eQTLs on autosomes. When we examined the effect of genetic control on sex-DEGs, we found genetic variants affecting the female-biased expression FCGR3A in NK cells (rs2099684) and ITGB2 in Monocytes (rs760462), both of which are associated with systemic lupus erythematosus. Our work reveals novel biases masked in bulk analyses and highlights sexually dimorphic genes and pathways at baseline.

immunology↗

Systematic comparison of Generative AI-Protein Models reveals fundamental differences between structural and sequence-based approaches.

Recent advances in artificial intelligence have led to the development of generative models for de novo protein design. We compared 13 state-of-the-art generative protein models, assessing their ability to produce feasible, diverse, and novel protein monomers. Structural diffusion models generally create designs with higher confidence in predicted structures and more biologically plausible energy distributions, but exhibit limited diversity and strong sequence biases. Conversely, protein language models generate more diverse and novel designs but with lower structural confidence. We also evaluated these models ability to generate unique proteins, conditionally based on the Tobacco Etch Virus (TEV) protease. Generative models were successful in producing functional enzymes, albeit with diminished activity compared to the wildtype TEV. Our systematic benchmarking provides a foundation for evaluating and selecting generative protein models, while highlighting the complementary strengths of different generative paradigms. This framework will facilitate an informed application of these tools for bio-medical engineering and design.

synthetic biology↗

Polymorphic tandem repeats shape single-cell gene expression across the immune landscape

Tandem repeats (TRs) - highly polymorphic, repetitive sequences across the human genome - are important regulators of gene expression but remain underexplored due to challenges in accurate genotyping and analysis1. Here, we generate new whole genome and single-cell RNA sequencing from >5.4 million blood-derived cells across 1,925 individuals in two cohorts [Cuomo et al., accompanying manuscript], and perform meta-analysis to characterize the impact of variation in >1.7 million TR loci on immune cell type-specific gene expression. We identify >69,000 single-cell expression TR loci (sc-eTRs), 30.7% of which are specific to one of 28 immune cell types, and reveal dynamic regulatory effects using cell-state inference. Matched single-cell ATAC sequencing profiles from >3.4 million nuclei in 922 individuals [Xue et al., accompanying manuscript]. uncover chromatin accessibility QTLs for nearly one-third of expression-associated TRs, supporting coordinated effects on cis-regulatory architecture. Fine-mapping implicates 1,490 TRs as candidate causal drivers of gene expression in 6.1% of tested genes, and colocalization analyses highlight >200 genes in which TRs likely mediate genetic associations with immune and hematological traits. Together, these results provide a genome-wide, multiomic view of TR-mediated regulation in the human immune system, establishing TRs as key contributors to cell type-specific regulatory variation and complex trait architecture.

genomics↗

Iso-Seq enables discovery of novel isoform variants in human retina at single cell resolution

Recent single cell transcriptomic profiling of the human retina provided important insights into the genetic signals in heterogeneous retinal cell populations that enable vision. However, conventional single cell RNAseq with 3 short-read sequencing is not suitable to identify isoform variants. Here we utilized Iso-Seq with full-length sequencing to profile the human retina at single cell resolution for isoform discovery. We generated a retina transcriptome dataset consisting of 25,302 nuclei from three donor retina, and detected 49,710 known transcripts and 241,949 novel transcripts across major retinal cell types. We surveyed the use of alternative promoters to drive transcript variant expression, and showed that 1-8% of genes utilized multiple promoters across major retinal cell types. Also, our results enabled gene expression profiling of novel transcript variants for inherited retinal disease (IRD) genes, and identified differential usage of exon splicing in major retinal cell types. Altogether, we generated a human retina transcriptome dataset at single cell resolution with full-length sequencing. Our study highlighted the potential of Iso-Seq to map the isoform diversity in the human retina, providing an expanded view of the complex transcriptomic landscape in the retina.

genomics↗

Genetic and cellular basis of impaired phagocytosis and photoreceptor degeneration in CLN3 disease.

PurposeCLN3 Batten disease (also known as Juvenile Neuronal Ceroid Lipofuscinosis; JNCL) is a lysosomal storage disorder that typically initiates with retinal degeneration but is followed by seizure onset, motor decline and premature death. Patient-derived CLN3 disease iPSC-RPE cells show defective phagocytosis of photoreceptor outer segments (POSs). Because modifier genes are implicated in CLN3 disease, our goal here was to investigate a direct link between CLN3 mutation and POS phagocytosis defect. MethodsIsogenic control and CLN3 mutant stem cell lines were generated by CRISPR-Cas9{-}mediated biallelic deletion of exons 7 and 8. A transgenic CLN3{Delta}7-8/{Delta}7-8 (CLN3) Yucatan miniswine was also used to study the impact of CLN3{Delta}7-8/{Delta}7-8 mutation on POS phagocytosis. POS phagocytosis by cultured RPE cells was analyzed by Western blotting and immunohistochemistry. Electroretinogram, optical coherence tomography and histological analysis of CLN3{Delta}7/8 and wild-type miniswine eyes were carried out at 6-, 36-, or 48-month age. ResultsCLN3{Delta}7-8/{Delta}7-8 RPE (CLN3 RPE) displayed reduced POS binding and consequently decreased uptake of POS compared to isogenic control RPE cells. Furthermore, wild-type miniswine RPE cells phagocytosed CLN3{Delta}7-8/{Delta}7-8 POS less efficiently than wild-type POS. Consistent with decreased POS phagocytosis, lipofuscin/autofluorescence was decreased in CLN3 miniswine RPE at 36 months-of-age and was followed by almost complete loss of photoreceptors at 48 months of age. ConclusionsCLN3{Delta}7-8/{Delta}7-8 mutation (that affects up to 85% patients) affects both RPE and POSs and leads to photoreceptor cell loss in CLN3 disease. Furthermore, both primary RPE dysfunction and mutant POS independently contribute to impaired POS phagocytosis in CLN3 disease.

cell biology↗

Spatial transcriptomic profiling of human retinoblastoma

Retinoblastoma (RB) represents one of the most prevalent intraocular cancers in children. Understanding the tumor heterogeneity in RB is important to design better targeted therapies. Here we used spatial transcriptomic to profile human retina and RB tumor to comprehensively dissect the spatial cell-cell communication networks. We found high intratumoral heterogeneity in RB, consisting of 10 transcriptionally distinct subpopulations with varying levels of proliferation capacity. Our results uncovered a complex architecture of the tumor microenvironment that predominantly consisted of cone precursors, as well as glial cells and cancer-associated fibroblasts. We delineated the cell trajectory underlying malignant progression of RB, and identified key signaling pathways driving genetic regulation across RB progression. We also explored the signaling pathways mediating cell-cell communications in RB subpopulations, and mapped the spatial networks of RB subpopulations and region neighbors. Altogether, we constructed the first spatial gene atlas for RB, which allowed us to characterize the transcriptomic landscape in spatially-resolved RB subpopulations, providing novel insights into the complex spatial communications involved in RB progression.

systems biology↗

Induced pluripotent stem cell derived pericytes respond to endogenous mediators of proliferation and contractility

BackgroundPericytes are multifunctional contractile cells that reside on capillaries. Pericytes are critical regulators of cerebral blood flow and blood-brain barrier function, and pericyte dysfunction may contribute to the pathophysiology of human neurological diseases including Alzheimers disease, multiple sclerosis, and stroke. Induced pluripotent stem cell (iPSC)-derived pericytes (iPericytes) are a promising tool for vascular research. However, it is unclear how iPericytes functionally compare to primary human brain vascular pericytes (HBVPs). We differentiated iPSCs into iPericytes of either the mesoderm or neural crest lineage using established protocols. We compared iPericyte and HBVP morphologies, quantified gene expression by qPCR and bulk RNA sequencing, and visualised pericyte protein markers by immunocytochemistry. To determine whether the gene expression of neural crest iPericytes, mesoderm iPericytes or HBVPs correlated with their functional characteristics in vitro, we quantified EdU incorporation following exposure to the key pericyte mitogen, platelet derived growth factor (PDGF)-BB and, contraction and relaxation in response to the vasoconstrictor endothelin-1 or vasodilator adenosine, respectively. iPericytes were morphologically similar to HBVPs and expressed canonical pericyte markers. However, iPericytes had 1864 differentially expressed genes compared to HBVPs, while there were 797 genes differentially expressed between neural crest and mesoderm iPericytes. Consistent with the ability of HBVPs to respond to PDGF-BB signalling, PDGF-BB enhanced and PDGF receptor-beta inhibitors impaired iPericyte proliferation. Administration of endothelin-1 led to iPericyte contraction and adenosine led to iPericyte relaxation, of a magnitude similar to the response evoked in HBVPs. We determined that neural crest iPericytes were less susceptible to PDGFR beta inhibition, but responded most robustly to vasoconstrictive meditators. iPericytes express pericyte-associated genes and proteins and, exhibit an appropriate physiological response upon exposure to a key endogenous mitogen or vasoactive mediators. Therefore, the generation of functional iPericytes would be suitable for use in future investigations exploring pericyte function or dysfunction in neurological diseases.

cell biology↗

Deep Learning-based Identification of Intraocular Pressure-Associated Genes Influencing Trabecular Meshwork Cell and Organelle Morphology

PURPOSEThe exact pathogenesis of primary open-angle glaucoma (POAG) is poorly understood. Genome-wide association studies (GWAS) have recently uncovered many loci associated with variation in intraocular pressure (IOP); a crucial risk factor for POAG. Artificial intelligence (AI) can be used to interrogate the effect of specific genetic knockouts on the morphology of trabecular meshwork cells (TMCs), the regulatory cells of IOP. METHODSSixty-two genes at fifty-five loci associated with IOP variation were knocked out in primary TMC lines. All cells underwent high-throughput microscopy imaging after being stained with a five-channel fluorescent cell staining protocol. A convolutional neural network (CNN) was trained to distinguish between gene knockout and normal control cell images. The area under the receiver operator curve (AUC) metric was used to quantify morphological variation in gene knockouts to identify potential pathological perturbations. RESULTSCells where RALGPS1 had been perturbed demonstrated the greatest morphological variation from normal TMCs (AUC 0.851, SD 0.030), followed by LTBP2 (AUC 0.846, SD 0.029) and BCAS3 (AUC 0.845, SD 0.020). Of seven multi-gene loci, five had statistically significant differences in AUC (p<0.05) between genes, allowing for pathological gene prioritisation. The mitochondrial channel most frequently showed the greatest degree of morphological variation (33.9% of cell lines). CONCLUSIONSWe demonstrate a robust method for functionally interrogating genome-wide association signals using high-throughput microscopy and AI. Genetic variations inducing marked morphological variation can be readily identified, allowing for the gene-based dissection of loci associated with complex traits.

genetics↗

Knockout of AMD-associated gene POLDIP2 reduces mitochondrial superoxide in human retinal pigment epithelial cells

Genetic and epidemiologic studies have significantly advanced our understanding of the genetic factors contributing to age-related macular degeneration (AMD). In particular, recent expression quantitative trait loci (eQTL) studies have highlighted POLDIP2 as a significant gene that confers risk of developing AMD. However, the role of POLDIP2 in retinal cells such as retinal pigment epithelium (RPE) and how it contributes to AMD pathology are unknown. Here we report the generation of a stable human RPE cell line with POLDIP2 knockout using CRISPR/Cas, providing an in vitro model to investigate the functions of POLDIP2. We conducted functional studies on the POLDIP2 knockout cell line and showed that they retained normal levels of cell proliferation, cell viability, phagocytosis and autophagy. Also, we performed RNA sequencing to profile the transcriptome of POLDIP2 knockout cells. Our results highlighted significant changes in genes involved in immune response, complement activation, oxidative damage and vascular development. We showed that loss of POLDIP2 causes a reduction in mitochondrial superoxide levels, which is consistent with the upregulation of the mitochondrial superoxide dismutase SOD2. In conclusion, this study demonstrates a novel link between POLDIP2 and SOD2, which supports a potential role of POLDIP2 in regulating oxidative stress in AMD pathology.

neuroscience↗

The VEGAS platform is not suitable for mammalian directed evolution

Directed evolution uses cycles of gene diversification and selection to generate proteins with novel properties. While traditionally directed evolution is performed in prokaryotic systems, recently a mammalian directed evolution system (viral evolution of genetically actuating sequences, or "VEGAS") has been described. Here we report that the VEGAS system has major limitations precluding its use for directed evolution. The primary technical issue with the VEGAS system is an immediate contamination with "cheater" particles that bypass directed evolution circuits. By sequencing we find these cheater particles contain Sindbis structural genes instead of the intended directed evolution target transgene. These cheaters outcompete the VEGAS transgenes within 2 rounds of transduction but cannot themselves activate synthetic circuits that drive expression of Sindbis structural genes, preventing directed evolution campaigns. Similar results have been obtained in independent labs. Taken together, the VEGAS system does not work as described and, without significant redesign to suppress cheaters, cannot be used for mammalian directed evolution campaigns.

synthetic biology↗

Lysosomal alterations and decreased electrophysiological activity in CLN3 disease (966 bp deletion, E295K) patient-derived cortical neurons

CLN3 disease is a lysosomal storage disorder associated with fatal neurodegeneration that is caused by mutations in CLN3. Most individuals with CLN3 disease carry at least one allele with a 966 bp deletion in CLN3 which results in the deletion of exons 7 and 8. There is a need for more physiologically relevant human cell-based CLN3 disease models to better understand the cellular changes during the disease process. Using CRISPR/Cas9, we corrected the 966 bp deletion mutation in human induced pluripotent stem cells (iPSCs) of a compound heterozygous patient (CLN3 {Delta} 966 bp and E295K). The isogenic deletion-corrected and unedited CLN3 patient iPSCs were used for disease modeling. iPSC-derived neurons carrying this particular CLN3 mutation (CLN3 neurons) had lower functional activity as recorded using microelectrode arrays for most of the culture period. Proteomics analysis showed downregulation of proteins related to axon guidance and endocytosis at day in vitro (DIV) 14 and 42 in CLN3 neurons. This was accompanied by an increase in lysosomal-related proteins in CLN3 neurons. Western blot analysis revealed hyperglycosylation of the lysosomal marker, Lysosome Associated Membrane Protein 1 (LAMP1) in CLN3 neurons at DIV 14, 28 and 42, which was not apparent in control neurons. Ultrastructural analysis of CLN3 neurons showed numerous membrane-bound vacuoles containing diverse types of storage material, ranging from curvilinear deposits, multilamellar structures to osmiophilic deposits. Our findings suggest alterations in lysosomal function and neurodevelopment involving axon guidance and synaptic transmission in CLN3-deficient neuronal derivatives, which could be potential targets for therapy.

neuroscience↗

propeller: testing for differences in cell type proportions in single cell data

MotivationSingle cell RNA Sequencing (scRNA-seq) has rapidly gained popularity over the last few years for profiling the transcriptomes of thousands to millions of single cells. This technology is now being used to analyse experiments with complex designs including biological replication. One question that can be asked from single cell experiments, which has been difficult to directly address with bulk RNA-seq data, is whether the cell type proportions are different between two or more experimental conditions. As well as gene expression changes, the relative depletion or enrichment of a particular cell type can be the functional consequence of disease or treatment. However, cell type proportions estimates from scRNA-seq data are variable and statistical methods that can correctly account for different sources of variability are needed to confidently identify statistically significant shifts in cell type composition between experimental conditions. ResultsWe have developed propeller, a robust and flexible method that leverages biological replication to find statistically significant differences in cell type proportions between groups. Using simulated cell type proportions data we show that propeller performs well under a variety of scenarios. We applied propeller to test for significant changes in proportions of cell types related to human heart development, ageing and COVID-19 disease severity. Availability and implementationThe propeller method is publicly available in the open source speckle R package (https://github.com/phipsonlab/speckle). All the analysis code for the paper is available at https://github.com/phipsonlab/propeller-paper-analysis/, and the associated analysis website is available at https://phipsonlab.github.io/propeller-paper-analysis/. ContactAlicia Oshlack: Alicia.Oshlack@petermac.org Belinda Phipson: phipson.b@wehi.edu.au Supplementary informationYes.

bioinformatics↗