Abstract
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.
Highlights
Single-cell RNA-sequencing has great potential to discover cell types, identify cell states, trace development lineages, and reconstruct the spatial organization of cells
Categorizing cell types comprising a specific organ or disease tissue is critical for comprehensive study of tissue development and function[1]
The data generated by these technologies enable us to quantify cell types, identify cell states, trace development lineages, and reconstruct the spatial organization of cells[18, 19]
Summary
Single-cell RNA-sequencing has great potential to discover cell types, identify cell states, trace development lineages, and reconstruct the spatial organization of cells. Dimensionality reduction projecting high-dimensional data into low-dimensional space (typically two or three dimensions) to visualize the cluster structures[27,28,29] and development trajectories[30,31,32,33] is commonly used Linear projection methods such as principal component analysis (PCA) typically cannot represent the complex structures of single-cell data in low dimensional spaces. Nonlinear dimension reduction, such as the t-distributed stochastic neighbor embedding algorithm (t-SNE)[34,35,36,37,38,39], has shown reasonable results for many applications and has been widely used in single-cell data processing[1, 40, 41]. We extensively tested our method on simulated data and several scRNA-seq datasets in both normal and malignant tissues to demonstrate the robustness of our method
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have