Interpretable dimensionality reduction of single cell transcriptome data with deep generative models

By Jiarui Ding, Anne Condon, Sohrab P. Shah. 2017

339 citations recorded by the source corpus.

Single-cell RNA-sequencing has great potential to discover cell types, identify cell states, trace development lineages, and reconstruct the spatial organization of cells. However, dimension reduction to interpret structure in single-cell sequencing data remains a challenge. Existing algorithms are either not able to uncover the clustering structures in the data or lose global information such as groups of clusters that are close to each other. We present a robust statistical model, scvis, to capture and visualize the low-dimensional structures in single-cell gene expression data. Simulation results demonstrate that low-dimensional representations learned by scvis preserve both the local and global neighbor structures in the data. In addition, scvis is robust to the number of data points and learns a probabilistic parametric mapping function to add new data points to an existing embedding. We then use scvis to analyze four single-cell RNA-sequencing datasets, exemplifying interpretable two-dimensional representations of the high-dimensional single-cell RNA-sequencing data. Although single-cell transcriptome data are increasingly available, their interpretation remains a challenge. Here, the authors present a dimensionality reduction approach that preserves both the local and global neighbourhood structures in the data thus enhancing its interpretability.

Subfields: Artificial Intelligence, Biophysics, Molecular Biology

How researchers see this paper

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

In Artificial Intelligence

In Biophysics

In Molecular Biology

Citations in the corpus

Machine Learning in Drug Discovery: A Review (2021, background): “For reducing high-dimensional into low dimensional form, Ding et al introduced probabilistic generative structure in gene expression of single-cell data accompanied by unpredictable estimations (Ding et al. 2018).”
Probabilistic harmonization and annotation of single‐cell transcriptomics data with deep generative models (2019, neutral): “One concern in applying methods based on neural networks (Ding et al , 2018; Wang & Gu, 2018; Amodio et al , 2019; Eraslan et al , 2019; Grønbech et al , 2020) in single-cell genomics and other domains is the robustness to hyperparameters choices (Hu & Greene, 2019).”
A deep generative model for single-cell RNA sequencing with application to detecting differentially expressed genes (2017, background): “Recent articles apply neural networks, but without an architecture based on biology [10, 11].”
Applications of machine learning in drug discovery and development (2019, background): “Ding et al. 75 developed a probabilistic generative model, scvis, to reduce the high-dimensional space to the low-dimensional structures in single-cell gene expression data with uncertainty estimates.”
Bayesian Inference for a Generative Model of Transcriptome Profiles from Single-cell RNA Sequencing (2018, methodology): “Another important feature of scVI and other methods based on neural networks [25, 26, 27, 28] is their scalability.”
Improved metagenome binning and assembly using deep variational autoencoders (2021, result): “…visualizing the two principal components of the input and latent encoding of the validation dataset, we found that the latent encoding seemed denoised and the clusters separated more clearly (Figures 2E, 2F), in line with the findings of a previous study using a VAE to enhance clustering 34 .”
A deep generative model for gene expression profiles from single-cell RNA sequencing (2017, background): “Recent articles apply neural networks, but without an architecture based on biology [10, 11].”
SOM-based embedding improves efficiency of high-dimensional cytometry data analysis (2019, methodology): “Other methods include SWNE, scvis, largevis [29, 6, 8], and the relatively new UMAP [15] that specically aims to provide beer, faster embedding than tSNE.”
Single-cell RNA-seq denoising using a deep count autoencoder (2018, background): “A number of recent studies describe applications of autoencoders in molecular biology 25 – 29 .”
Current best practices in single‐cell RNA‐seq analysis: a tutorial (2019, positive): “First applications to scRNA-seq are starting to emerge from dimensionality reduction to denoising (e.g. scVis: Ding et al , 2018; scGen: preprint: Lotfollahi et al , 2018; DCA: Eraslan et al , 2019).”