Identifying Cell Subpopulations and Their Genetic Drivers from Single-Cell RNA-Seq Data Using a Biclustering Approach

Abstract

Single-cell RNA-Seq (scRNA-Seq) has attracted much attention recently because it allows unprecedented resolution into cellular activity; the technology, therefore, has been widely applied in studying cell heterogeneity such as the heterogeneity among embryonic cells at varied developmental stages or cells of different cancer types or subtypes. A pertinent question in such analyses is to identify cell subpopulations as well as their associated genetic drivers. Consequently, a multitude of approaches have been developed for clustering or biclustering analysis of scRNA-Seq data. In this article, we present a fast and simple iterative biclustering approach called "BiSNN-Walk" based on the existing SNN-Cliq algorithm. One of BiSNN-Walk's differentiating features is that it returns a ranked list of clusters, which may serve as an indicator of a cluster's reliability. Another important feature is that BiSNN-Walk ranks genes in a gene cluster according to their level of affiliation to the associated cell cluster, making the result more biologically interpretable. We also introduce an entropy-based measure for choosing a highly clusterable similarity matrix as our starting point among a wide selection to facilitate the efficient operation of our algorithm. We applied BiSNN-Walk to three large scRNA-Seq studies, where we demonstrated that BiSNN-Walk was able to retain and sometimes improve the cell clustering ability of SNN-Cliq. We were able to obtain biologically sensible gene clusters in terms of GO term enrichment. In addition, we saw that there was significant overlap in top characteristic genes for clusters corresponding to similar cell states, further demonstrating the fidelity of our gene clusters.

Bioinformatic Methods for Identifying Differentially Abundant Subpopulations in scRNA-seq Data

“…[40] Recent tools such as DivBiclust, PanoView, and scziDesk have been developed in order to make such analysis possible on larger datasets through biclustering or iterative clustering that can scale with dataset size reducing the need for dimensionality reduction. [41][42][43][44] An even newer unsupervised clustering approach called scGAC also seeks to analyze high dimensional and sparse datasets. The method utilizes latent relationship information across cells to graphically obtain cell clusters.…”

Network Modeling in Biology: Statistical Methods for Gene and Brain Networks

“…While it is useful for de novo discovery of new cell types and subtypes, unsupervised learning depends on many user-specific inputs, including which clustering algorithm to use (e.g., K-means clustering, hierarchical clustering, density-based clustering or graph-based clustering), the type of similarity or distance metric between two cells, and the number of clusters, which is a key parameter needed for many clustering algorithms. Taking into account the distinct features of scRNA-seq data, multiple cell clustering algorithms have been developed, including SNN-Cliq, which does not use conventional similarity measures but leverages the ranking of cells to construct a cell-cell graph for identifying cell clusters [244]; BiSNN-Walk, which extends SNN-Cliq and uses an iterative biclustering approach to return a ranked list of cell clusters, each associated with a set of ranked genes based on their levels of affiliation with the cluster [192]; CIDR, the first clustering method that incorporates imputation of dropout gene expression levels [125]; SC3, a widelyused ensemble method that combines multiple clustering algorithms [106]; and Seurat, which identifies cell clusters based on a shared nearest neighbor (SNN) clustering algorithm [184]. In addition to commonly used similarity metric including the Pearson correlation, Spearman correlation, Euclidean distance, other cell similarity measures can be found in, for example, [91,186].…”

It is time to apply biclustering: a comprehensive review of biclustering applications in biological and biomedical data

“…The recent advent of scRNA-seq technology has enabled researchers to study heterogeneity between individual cells and define cell type a based solely on its transcriptome [132]. Using biclustering, researchers can not only group cells into subpopulations but also identify biologically important gene signatures for each class simultaneously [95,139]. For example, Zeisel et al [95] recently classified single cells from the brain through biclustering, which identified numerous marker genes and highly restricted expression patterns of transcription factors for cell types.…”