The art of using t-SNE for single-cell transcriptomics

Искусство применения t-SNE для транскриптомики отдельных клеток
Dmitry Kobak, Philipp Berens
2019-11-28

PCA initializationdimensionality reductionmulti-scale similarity kernelssingle-cell transcriptomicst-SNE
Single-cell transcriptomics yields ever growing data sets containing RNA expression levels for thousands of genes from up to millions of cells. Common data analysis pipelines include a dimensionality reduction step for visualising the data in two dimensions, most frequently performed using t-distributed stochastic neighbour embedding (t-SNE). It excels at revealing local structure in high-dimensional data, but naive applications often suffer from severe shortcomings, e.g. the global structure of the data is not represented accurately. Here we describe how to circumvent such pitfalls, and develop a protocol for creating more faithful t-SNE visualisations. It includes PCA initialisation, a high learning rate, and multi-scale similarity kernels; for very large data sets, we additionally use exaggeration and downsampling-based initialisation. We use published single-cell RNA-seq data sets to demonstrate that this protocol yields superior results compared to the naive application of t-SNE.
1
Experiments on published single-cell RNA-seq datasets show that the protocol produces superior visualizations compared with naive t-SNE application.
2
For very large datasets, exaggeration and downsampling-based initialization further support more faithful t-SNE representations.
3
Naive t-SNE applications in single-cell transcriptomics can substantially misrepresent the global structure of high-dimensional expression data.
4
The proposed protocol improves t-SNE visualizations through PCA initialization, a high learning rate, and multi-scale similarity kernels.

single-cell transcriptomics data sets containing gene expression profiles from individual cells

the fidelity and representation of local and global data structure in t-SNE visualisations, including the effects of protocol choices and pitfalls

Publication Details
Publication Date
2019-11-28
Journal
Publisher
ISSN
Cited by
1180
Access Type
Author Information
Authors
Dmitry Kobak
Philipp Berens
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%