Geometric Sketching Compactly Summarizes the Single-Cell Transcriptomic Landscape

Hie, Brian; Cho, Hyunghoon; DeMeo, Benjamin; Bryson, Bryan; Berger, Bonnie

doi:10.1016/j.cels.2019.05.003

Cited by 115 publications

(107 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the computational complexity is O (kn 2 ). scMC can analyze large-scale datasets, e.g., 100 K cells within 2 h. To reduce the run time, users can use fewer confident cells by adjusting the quantile cutoff parameter (e.g., from 0.75 to 0.9) or perform subsampling of single-cell datasets using a "geometric sketching" approach [40], which allows to maintain the transcriptomic heterogeneity within a dataset with a smaller subset of cells. More efficient methods and packages for calculating eigenvalues will significantly enhance the computational speed of scMC.…”

Section: Discussionmentioning

confidence: 99%

scMC learns biological variation through the alignment of multiple single-cell genomics datasets

2021

View full text Add to dashboard Cite

Distinguishing biological from technical variation is crucial when integrating and comparing single-cell genomics datasets across different experiments. Existing methods lack the capability in explicitly distinguishing these two variations, often leading to the removal of both variations. Here, we present an integration method scMC to remove the technical variation while preserving the intrinsic biological variation. scMC learns biological variation via variance analysis to subtract technical variation inferred in an unsupervised manner. Application of scMC to both simulated and real datasets from single-cell RNA-seq and ATAC-seq experiments demonstrates its capability of detecting context-shared and context-specific biological signals via accurate alignment.

show abstract

Section: Discussionmentioning

confidence: 99%

scMC learns biological variation through the alignment of multiple single-cell genomics datasets

2021

View full text Add to dashboard Cite

show abstract

“…In the field of flow cytometry, this has resulted in the generation of algorithms such as flowSOM (26), which implements self-organizing maps to reduce the number of observations and effectively decrease computational costs, and PhenoGraph (25), which approximates the data using KNN-graphs and subsequently clusters using the Louvain algorithm allowing the analysis of thousands of individual cells. Recently, single cell RNAseq has become increasingly popular, spurring the development of even more efficient clustering methods such as the Leiden clustering (21), dimensionality reduction techniques such as UMAP (22) and subsampling methods that preserve the topology of the original data, such as geometric sketching (20). Here, we combined these methods to perform an initial subsampling of the flow cytometric data, effectively reducing the computational cost, followed by the generation of a KNN-graph and Leiden clustering.…”

Section: Discussionmentioning

confidence: 99%

“…In addition, we applied a computational approach, based on established single cell RNAseq pipelines (19), which consists of two steps. First, in order to improve computational throughput, compensated and biexponentially transformed data is subsampled and pooled, either in a random manner or by using geometric sketching (20) which allows for enrichment of rare subpopulations, or these subsampling methods can be combined as was done in all the following analyses. Subsequently, a k-nearest neighbor (KNN) graph was constructed and Leiden clustering was performed (21) to allow identification of populations within the data, which in turn were visualized using an UMAP manifold (22) ( Figure 1C).…”

Section: The Human Postnatal Thymus Is Characterized By An Increased mentioning

confidence: 99%

Conventional and Computational Flow Cytometry Analyses Reveal Sustained Human Intrathymic T Cell Development From Birth Until Puberty

et al. 2020

View full text Add to dashboard Cite

The thymus is the organ where subsets of mature T cells are generated which subsequently egress to function as central mediators in the immune system. While continuously generating T cells even into adulthood, the thymus does undergo involution during life. This is characterized by an initial rapid decrease in thymic cellularity during early life and by a second age-dependent decline in adulthood. The thymic cellularity of neonates remains low during the first month after birth and the tissue reaches a maximum in cellularity at 6 months of age. In order to study the effect that this first phase of thymic involution has on thymic immune subset frequencies, we performed multi-color flow cytometry on thymic samples collected from birth to 14 years of age. In consideration of the inherent limitations posed by conventional flow cytometry analysis, we established a novel computational analysis pipeline that is adapted from single-cell transcriptome sequencing data analysis. This allowed us to overcome technical effects by batch correction, analyze multiple samples simultaneously, limit computational cost by subsampling, and to rely on KNN-graphs for graph-based clustering. As a result, we successfully identified rare, distinct and gradually developing immune subsets within the human thymus tissues. Although the thymus undergoes early involution from infanthood onwards, our data suggests that this does not affect human T-cell development as we did not observe significant alterations in the proportions of T-lineage developmental intermediates from birth to puberty. Thus, in addition to providing an interesting novel strategy to analyze conventional flow cytometry data for the thymus, our work shows that the early phase of human thymic involution mainly limits the overall T cell output since no obvious changes in thymocyte subsets could be observed.

show abstract

“…In a future study, to resolve the limitation of D-EE on huge-scale data computation, we can accelerate D-EE by adopting either the fast Fourier transform as used in FIt-SNE, or adopting the state-of-the-art neural network framework used by net-SNE [ 28 ]. On the other hand, because a huge-scale single-cell dataset can be highly redundant, we can also select a subset of informative samples using an advanced geometric sketching tool [ 29 ] prior to application of D-EE.…”

Section: Discussionmentioning

confidence: 99%

D-EE: Distributed software for visualizing intrinsic structure of large-scale single-cell data

Huang

Wan

2020

GigaScience

View full text Add to dashboard Cite

Background Dimensionality reduction and visualization play vital roles in single-cell RNA sequencing (scRNA-seq) data analysis. While they have been extensively studied, state-of-the-art dimensionality reduction algorithms are often unable to preserve the global structures underlying data. Elastic embedding (EE), a nonlinear dimensionality reduction method, has shown promise in revealing low-dimensional intrinsic local and global data structure. However, the current implementation of the EE algorithm lacks scalability to large-scale scRNA-seq data. Results We present a distributed optimization implementation of the EE algorithm, termed distributed elastic embedding (D-EE). D-EE reveals the low-dimensional intrinsic structures of data with accuracy equal to that of elastic embedding, and it is scalable to large-scale scRNA-seq data. It leverages distributed storage and distributed computation, achieving memory efficiency and high-performance computing simultaneously. In addition, an extended version of D-EE, termed distributed optimization implementation of time-series elastic embedding (D-TSEE), enables the user to visualize large-scale time-series scRNA-seq data by incorporating experimentally temporal information. Results with large-scale scRNA-seq data indicate that D-TSEE can uncover oscillatory gene expression patterns by using experimentally temporal information. Conclusions D-EE is a distributed dimensionality reduction and visualization tool. Its distributed storage and distributed computation technique allow us to efficiently analyze large-scale single-cell data at the cost of constant time speedup. The source code for D-EE algorithm based on C and MPI tailored to a high-performance computing cluster is available at https://github.com/ShaokunAn/D-EE.

show abstract

Geometric Sketching Compactly Summarizes the Single-Cell Transcriptomic Landscape

Cited by 115 publications

References 54 publications

scMC learns biological variation through the alignment of multiple single-cell genomics datasets

scMC learns biological variation through the alignment of multiple single-cell genomics datasets

Conventional and Computational Flow Cytometry Analyses Reveal Sustained Human Intrathymic T Cell Development From Birth Until Puberty

D-EE: Distributed software for visualizing intrinsic structure of large-scale single-cell data

Contact Info

Product

Resources

About