Inference after latent variable estimation for single-cell RNA sequencing data

Neufeld, Anna; Gao, Lucy L.; Popp, Joshua; Battle, Alexis; Witten, Daniela

doi:10.48550/arxiv.2207.00554

Cited by 4 publications

(6 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We first, created a new open source python-based bootstrapping implementation of the count-splitting method 8 , employing it to create three independent measures for training-, validation-, and test-sets that host each cell and gene in every split. Note that this differs from standard machine learning (ML) practice in which separate samples (cells in scRNAseq) would be used for training, validation, and test, looking for replicated characteristics in different populations.…”

Section: Resultsmentioning

confidence: 99%

“…Here we present a count-splitting approach inspired by Neufeld et al 8 , that uses a training-set for topology building and clustering, a validation-set to optimize the training-clusters (or other topology-associated analysis), and a final test-set for DEG analysis. These splits enabled our creation of a new cluster validation algorithm and also allowed us to develop the first self-supervised benchmark of scRNAseq analysis pipelines.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Self-supervised Benchmarking for scRNAseq Clustering

Tyler

Guccione

Schadt

2023

Preprint

View full text Add to dashboard Cite

Interpretation of single cell RNAseq (scRNAseq) data are typically built upon clustering results and/or cell-cell topologies. However, the validation process is often exclusively left to bench biologists, which can take years and tens of thousands of dollars. Furthermore, a lack of objective ground-truth labels in complex biological datasets, has resulted in difficulties when benchmarking single cell analysis methods. Here, we address these gaps with count splitting, creating a cluster validation algorithm, accounting for Poisson sampling noise, and benchmark 120 pipelines using an independent test-set for ground-truth assessment, thus enabling the firstself-supervisedbenchmark. Anti-correlation-based feature selection paired with locally weighted Louvain modularity on the Euclidean distance of 50 principal-components with cluster-validation showed the best performance of all tested pipelines for scRNAseq clustering, yielding reproducible biologically meaningful populations. These new approaches enabled the discovery of a novel metabolic gene signature associated with hepatocellular carcinoma survival time.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Self-supervised Benchmarking for scRNAseq Clustering

Tyler

Guccione

Schadt

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…To mitigate double-dipping, a Poisson/negative binomial sampling of the original data into independent partitions of the same dimensions was introduced in an approach called count splitting [Neufeld et al, 2022[Neufeld et al, , 2023. The first partition can be used for latent-variable estimation and the second partition for inference.…”

Section: Discussionmentioning

confidence: 99%

Tree-based differential testing using inferential uncertainty for RNA-Seq

Singh,

Wu,

Fan

et al. 2023

Preprint

View full text Add to dashboard Cite

Identifying differentially expressed transcripts poses a crucial yet challenging problem in transcriptomics. Substantial uncertainty is associated with the abundance estimates of certain transcripts which, if ignored, can lead to the exaggeration of false positives and, if included, may lead to reduced power. For a given set of RNA-Seq samples,TreeTerminusarranges transcripts in a hierarchical tree structure that encodes different layers of resolution for interpretation of the abundance of transcriptional groups, with uncertainty generally decreasing as one ascends the tree from the leaves. We introducetrenDi, which utilizes the tree structure fromTreeTerminusfor differential testing. The candidate nodes are determined in a data-driven manner to maximize the signal that can be extracted from the data while controlling for the uncertainty associated with estimating the transcript abundances. The identified candidate nodes can include transcripts and inner nodes, with no two nodes having an ancestor/descendant relationship. We evaluated our method on both simulated and experimental datasets, comparing its performance with other tree-based differential methods as well as with uncertainty-aware differential transcript/gene expression methods. Our method detects inner nodes that show a strong signal for differential expression, which would have been overlooked when analyzing the transcripts alone.

show abstract

“…To characterize previously unknown cell types, automatic selection of signature genes for each cluster is often achieved through differential expression (DE) testing [24]. For this task, BacSC provides capabilities for DE testing that takes the recently popularized problem of "double dipping" for DE testing of cell types into account [29,45,46]. In short, using the same information (gene expression) to define a clustering as well as the subsequently determining DE genes to characterize these clusters results in an inflated false discovery rate (FDR).…”

Section: Description Of the Bacsc Pipelinementioning

confidence: 99%

BacSC: A general workflow for bacterial single-cell RNA sequencing data analysis

Ostner,

Kirk,

Olayo-Alarcon

et al. 2024

Preprint

View full text Add to dashboard Cite

Bacterial single-cell RNA sequencing has the potential to elucidate within-population heterogeneity of prokaryotes, as well as their interaction with host systems. Despite conceptual similarities, the statistical properties of bacterial single-cell datasets are highly dependent on the protocol, making proper processing essential to tap their full potential. We present BacSC, a fully data-driven computational pipeline that processes bacterial single-cell data without requiring manual intervention. BacSC performs data-adaptive quality control and variance stabilization, selects suitable parameters for dimension reduction, neighborhood embedding, and clustering, and provides false discovery rate control in differential gene expression testing. We validated BacSC on a broad selection of bacterial single-cell datasets spanning multiple protocols and species. Here, BacSC detected subpopulations inKlebsiella pneumoniae, found matching structures ofPseudomonas aeruginosaunder regular and low-iron conditions, and better represented subpopulation dynamics ofBacillus subtilis. BacSC thus simplifies statistical processing of bacterial single-cell data and reduces the danger of incorrect processing.

show abstract

Inference after latent variable estimation for single-cell RNA sequencing data

Cited by 4 publications

References 18 publications

Self-supervised Benchmarking for scRNAseq Clustering

Self-supervised Benchmarking for scRNAseq Clustering

Tree-based differential testing using inferential uncertainty for RNA-Seq

BacSC: A general workflow for bacterial single-cell RNA sequencing data analysis

Contact Info

Product

Resources

About