Sub-Cluster Identification through Semi-Supervised Optimization of Rare-cell Silhouettes (SCISSORS) in Single-Cell Sequencing

Abstract

Single-cell RNA-sequencing (scRNA-seq) has enabled the molecular profiling of thousands to millions of cells simultaneously in biologically heterogenous samples. Currently, common practice in scRNA-seq is to determine cell type labels through unsupervised clustering and the examination of cluster-specific genes. However, even small differences in analysis and parameter choice can greatly alter clustering solutions and thus impose great influence on which cell types are identified. Existing methods largely focus on determining the optimal number of robust clusters, which is not favorable for identifying cells of extremely low abundance due to their subtle contributions towards overall patterns of gene expression. Here we present a carefully designed framework, SCISSORS, which accurately profiles subclusters within major cluster(s) for the identification of rare cell types in scRNA-seq data. SCISSORS employs silhouette scoring for the estimation of heterogeneity of clusters and reveals rare cells in heterogenous clusters by implementing a multi-step, semi-supervised reclustering process. Additionally, SCISSORS provides a method for the identification of marker genes of rare cells, which may be used for further study. SCISSORS is wrapped around the popular Seurat R package and can be easily integrated into existing Seurat pipelines. SCISSORS, including source code and vignettes for two example datasets, is freely available at https://github.com/jrleary/SCISSORS.

Systems Level Identification of a Matrisome-Associated Macrophage Polarization State in Multi-Organ Fibrosis

“…To derive sub- clusters within the macrophage population in an unbiased manner, Louvain cluster analysis was performed using Seurat’s FindCluster function. Specifically, the analysis was performed multiple times with different cluster resolutions (from .2 to 1 in increments of .05), and the cluster result at each resolution was assessed by the Silhouette score [116]. Briefly, Silhouette score evaluates the goodness of clustering by measuring for every cell the similarity with cells in the same cluster as compared to the similarity to cells in other clusters.…”

“…It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 20, 2022. ; https://doi.org/10.1101/2022.12.20.521178 doi: bioRxiv preprint performed using Seurat's FindCluster function. Specifically, the analysis was performed multiple times with different cluster resolutions (from .2 to 1 in increments of .05), and the cluster result at each resolution was assessed by the Silhouette score [116]. Briefly, Silhouette score evaluates the goodness of clustering by measuring for every cell the similarity with cells in the same cluster as compared to the similarity to cells in other clusters.…”

Determination of permissive and restraining cancer-associated fibroblast (DeCAF) subtypes

“…Therefore, we set out to translate exemplar or marker genes derived from scRNAseq for use in bulk transcriptomics data. We previously showed that the basal-like and classical cell-subpopulation-specific marker genes derived from scRNAseq may be translated to cluster bulk RNAseq of PDAC samples, and highly recapitulated the clinically validated PDAC tumor subtypes 26 . These marker genes were identified using a previously developed tool, SCISSORS, which sensitively identifies rare cell subpopulations as low as 0.092% of the cell population and accurately identifies marker genes of high specificity.…”

“…SCISSORS includes a carefully designed function, which is a two-step method for the identification of highly cell subpopulation specific genes 26 . Briefly, SCISSORS first derives a candidate gene set by comparing the cell subpopulation of interest to the most related cell subpopulation; then the highly expressed genes from other unrelated cell types are removed from this candidate gene set.…”

“…Using SCISSORS, a method that we previously developed to sensitively cluster rare cells and identify highly cell-subpopulation-specific marker genes in scRNAseq, we identify uniquely expressed marker genes for CAF clusters that robustly translate to bulk RNAseq 26 . We show that these marker genes are specific to fibroblasts and are not confounded by expression in immune cells.…”