Visualization and analysis of single-cell RNA-seq data by kernel-based similarity learning

By Junjie Zhu, Emma Pierson, Daniele Ramazzotti, Serafim Batzoglou. Nature Methods, 2016

650 citations recorded by the source corpus.

We present single-cell interpretation via multikernel learning (SIMLR), an analytic framework and software which learns a similarity measure from single-cell RNA-seq data in order to perform dimension reduction, clustering and visualization. On seven published data sets, we benchmark SIMLR against state-of-the-art methods. We show that SIMLR is scalable and greatly enhances clustering performance while improving the visualization and interpretability of single-cell sequencing data.

Subfields: Biophysics, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/77e2de484be3ed27f3f2a02a4e01b3fc964d65db

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biophysics

In Molecular Biology

Citations in the corpus

Deep Generative Modeling for Single-cell Transcriptomics (2018, methodology): “For instance, SIMLR [10] ﬁ ts a cell-cell similarity matrix, under the assumption that this matrix has a block structure with a ﬁ xed number of clusters.”
Graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells (2017, background): “Any other distance measure, for example, the kernel-based measure of Reference [45], would also be a viable option.”
Integration strategies of multi-omics data for machine learning analysis (2021, background): “On the other hand, SIMLR [93] and its multi-omics extension CIMLR [94] compute several gaussian kernels with varying hyperparameters for each omics and assume that the global kernel matrix better reﬂects similarity between samples and naturally reﬂects possible clusters.”
Bayesian Inference for a Generative Model of Transcriptome Profiles from Single-cell RNA Sequencing (2018, methodology): “Next, we benchmark scVI with SIMLR [12], a method that couples clustering with learning of a cell-cell similarity matrix.”
A general and flexible method for signal extraction from single-cell RNA-seq data (2017, neutral): “[48] have proposed a novel visualization algorithm that can account for zero inflation and showed improvement over t-SNE.”
DrImpute: imputing dropout events in single cell RNA sequencing data (2017, methodology): “Moreover, our experience with DrImpute suggested that the range of k needed to be no less than the real number of cell clusters (though unlike other methods such as scImpute, pcaReduce or SIMLR, the exact number of expected clusters do not need to be specified) [23, 25, 48].”
Challenges in unsupervised clustering of single-cell RNA-seq data (2019, methodology): “…leads to the greatest reduction of the cost function. greedy algorithms are often fast, but they may fail to find the optimal solution. www.nature.com/nrg R e v i e w s detection to identify rare cell types, whereas SIMLR 24 adapts k means by simultaneously training a custom distance measure.”
Interpretable dimensionality reduction of single cell transcriptome data with deep generative models (2017, methodology): “In addition, although the learned similarity matrix could help clustering analyses, it may distort the manifold structure as demonstrated in the t-SNE plots on the learned similarity matrix 54 because the SIMLR objective function encourages forming clusters.”
SingleCellSignalR: inference of intercellular networks from single-cell transcriptomics (2019, methodology): “Clustering can be obtained either by chaining principal component analysis for dimension reduction and K-means (35), or by using the advanced SIMLR model (36).”
Current best practices in single‐cell RNA‐seq analysis: a tutorial (2019, methodology): “Alternatives to the Web-based GUI platforms are packages such as FASTGenomics (preprint: Scholz et al , 2018), iSEE (Rue-Albrecht et al , 2018), IS-CellR (Patel, 2018), and Granatum run on a local server.”
Integrating single-cell transcriptomic data across different conditions, technologies, and species (2018, methodology): “For example, zero-inflated differential expression tests have been tailored to scRNA-seq data to identify changes within a single-cell type 16,17 , and clustering approaches 18–23 can detect proportional shifts across conditions if cell types are conserved.”
Clustering single-cell RNA-seq data with a model-based deep learning approach (2019, neutral): “Wang and colleagues employ multi-kernel learning (SIMLR) for single-cell interpretation 15 .”
Using neural networks for reducing the dimensions of single-cell RNA-Seq data (2017, methodology): “(B) 2D PCA for SIMLR-transformed data.”
Generalizable and Scalable Visualization of Single-Cell Data Using Neural Networks (2018, methodology): “, dropout events where nonzero expression levels are missed as zero) have also been proposed (Pierson and Yau, 2015; Wang et al., 2017).”
Revealing the vectors of cellular identity with single-cell genomics (2016, background): “A very recent approach, SIMLR 119 , learns a cell-to-cell similarity matrix S that is based on a combination of multiple kernels (rather than a predefined distance metric) by assuming that there exist C discernible cell clusters and exploiting the nearly block-diagonal structure they induce on S .”
Information Constraints on Auto-Encoding Variational Bayes (2018, background): “A common application for scRNA-seq is discovering cell types, which can be be done without correcting for the alignment errors [37].”