scCODA: A Bayesian model for compositional single-cell data analysis

Büttner, Maren; Ostner, Johannes; Müller, Cl.; Theis, Fabian J.; Schubert, Benjamin

doi:10.1101/2020.12.14.422688

Cited by 36 publications

(57 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Popular choices include picking one of the p features or the (geometric) mean over multiple or all groups (Fernandes et al, 2014). Following the scCODA model, we pick a single reference feature prior to analysis (Büttner et al, 2020). Technically, this is achieved by choosing one featurep that is set to be unchanged by all covariates.…”

Section: Reference Featurementioning

confidence: 99%

“…tascCODA (Kumar et al (2019)), numpy=1.19.5, scanpy=1.8.1 (Wolf et al (2018)), toytree=2.0.1, and sccoda=0.1.4 (Büttner et al (2020)).…”

Section: Computational Aspectsmentioning

confidence: 99%

“…To test the performance of tascCODA in a differential abundance testing scenario, we generated compositional datasets with an underlying tree structure and compared how well several models could detect the changes introduced by a binary covariate. For compositional models that do not account for the tree structure, we used the state-of-the art methods ANCOM-BC (Lin and Peddada ( 2020)), ANCOM (Mandal et al (2015)), and ALDEx2 (Fernandes et al (2014)) from the field of microbiome data analysis, as well as scCODA (Büttner et al, 2020) from scRNA-seq analysis. Based on the recommendations by Aitchison (1982), we also analyzed the data with the additive log-ratio (ALR) transformation in combination with t-or Wilcoxon rank-sum tests.…”

Section: Model Comparisonmentioning

confidence: 99%

“…In its present form, the Bayesian model behind tascCODA is ideally suited for data sets of moderate dimensionality, typically p < 100, yet can handle extremely small sample sizes n. Since amplicon datasets are usually high-dimensional in the number of taxa and exhibit high overdispersion and excess number of zeros, we focus on the analysis of genus-level microbiome data. In the context of cell type compositional data, on the other hand, often only very few replicate samples are available (Büttner et al, 2020). Here, tascCODA can leverage well-calibrated prior information to operate in low-sample regimes where frequentist methods likely fail.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

tascCODA: Bayesian tree-aggregated analysis of compositional amplicon and single-cell data

Ostner

Carcy

Müller

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Accurate generative statistical modeling of count data is of critical relevance for the analysis of biological datasets from high-throughput sequencing technologies. Important instances include the modeling of microbiome compositions from amplicon sequencing surveys and the analysis of cell type compositions derived from single-cell RNA sequencing. Microbial and cell type abundance data share remarkably similar statistical features, including their inherent compositionality and a natural hierarchical ordering of the individual components from taxonomic or cell lineage tree information, respectively. To this end, we introduce a Bayesian model for tree-aggregated amplicon and single-cell compositional data analysis tascCODA that seamlessly integrates hierarchical information and experimental covariate data into the generative modeling of compositional count data. By combining latent parameters based on the tree structure with spike-and-slab Lasso penalization, tascCODA can determine covariate effects across different levels of the population hierarchy in a data-driven parsimonious way. In the context of differential abundance testing, we validate tascCODA's excellent performance on a comprehensive set of synthetic benchmark scenarios. Our analyses on human single-cell RNA-seq data from ulcerative colitis patients and amplicon data from patients with irritable bowel syndrome, respectively, identified aggregated cell type and taxon compositional changes that were more predictive and parsimonious than those proposed by other schemes. We posit that tascCODA constitutes a valuable addition to the growing statistical toolbox for generative modeling and analysis of compositional changes in microbial or cell population data.

show abstract

Section: Reference Featurementioning

confidence: 99%

“…tascCODA (Kumar et al (2019)), numpy=1.19.5, scanpy=1.8.1 (Wolf et al (2018)), toytree=2.0.1, and sccoda=0.1.4 (Büttner et al (2020)).…”

Section: Computational Aspectsmentioning

confidence: 99%

Section: Model Comparisonmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

tascCODA: Bayesian tree-aggregated analysis of compositional amplicon and single-cell data

Ostner

Carcy

Müller

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…The proportion of each cell type was compared across samples using scCODA (version 0.1.1). 65 Briefly, we constructed a Markov-Chain Monte Carlo model with Hamiltonian Monte Carlo sampling using cell type proportions between conditions (non-COVID-19 reference control versus COVID-19, or COVID-19 microthrombi-positive versus COVID-19 microthrombi-negative).…”

Section: Compositional Analysesmentioning

confidence: 99%

Molecular Pathophysiology of Cardiac Injury and Cardiac Microthrombi in Fatal COVID-19: Insights from Clinico-histopathologic and Single Nuclei RNA Sequencing Analyses

Fukuma

Hulke

Brener

et al. 2021

Preprint

View full text Add to dashboard Cite

Cardiac injury is strongly associated with critical COVID-19, yet its etiology remains debated. To elucidate the pathogenic mechanisms of COVID-19-associated cardiac injury, we conducted a single-center prospective cohort study of 69 COVID-19 decedents. Of six cardiac histopathologic features, microthrombi was the most commonly detected (n=48, 70%). We tested associations of cardiac microthrombi with biomarkers of inflammation, cardiac injury, and fibrinolysis and with in-hospital antiplatelet therapy, therapeutic anticoagulation, and corticosteroid treatment, while adjusting for multiple clinical factors, including COVID-19 therapies. Higher peak ESR and CRP during hospitalization were independently associated with higher odds of microthrombi. Using single nuclei RNA-sequence analysis, we discovered an enrichment of pro-thrombotic/anti-fibrinolytic, extracellular matrix remodeling, and immune-potentiating signaling amongst cardiac fibroblasts in microthrombi-positive COVID-19 hearts relative to microthrombi-negative COVID-19 and non-COVID-19 reference hearts. Our cumulative findings identify the specific transcriptomic changes in cardiac fibroblasts as salient features of COVID-19-associated cardiac microthrombi.

show abstract

Primary cilia and SHH signaling impairments in human and mouse models of Parkinson’s disease

et al. 2022

Self Cite

View full text Add to dashboard Cite

Parkinson’s disease (PD) as a progressive neurodegenerative disorder arises from multiple genetic and environmental factors. However, underlying pathological mechanisms remain poorly understood. Using multiplexed single-cell transcriptomics, we analyze human neural precursor cells (hNPCs) from sporadic PD (sPD) patients. Alterations in gene expression appear in pathways related to primary cilia (PC). Accordingly, in these hiPSC-derived hNPCs and neurons, we observe a shortening of PC. Additionally, we detect a shortening of PC in PINK1-deficient human cellular and mouse models of familial PD. Furthermore, in sPD models, the shortening of PC is accompanied by increased Sonic Hedgehog (SHH) signal transduction. Inhibition of this pathway rescues the alterations in PC morphology and mitochondrial dysfunction. Thus, increased SHH activity due to ciliary dysfunction may be required for the development of pathoetiological phenotypes observed in sPD like mitochondrial dysfunction. Inhibiting overactive SHH signaling may be a potential neuroprotective therapy for sPD.

show abstract

scCODA: A Bayesian model for compositional single-cell data analysis

Cited by 36 publications

References 32 publications

tascCODA: Bayesian tree-aggregated analysis of compositional amplicon and single-cell data

tascCODA: Bayesian tree-aggregated analysis of compositional amplicon and single-cell data

Molecular Pathophysiology of Cardiac Injury and Cardiac Microthrombi in Fatal COVID-19: Insights from Clinico-histopathologic and Single Nuclei RNA Sequencing Analyses

Primary cilia and SHH signaling impairments in human and mouse models of Parkinson’s disease

Contact Info

Product

Resources

About