Abstract
Recent advances in single-cell technologies have enabled high-resolution characterization of tissue and cancer compositions. Although numerous tools for dimension reduction and clustering are available for single-cell data analyses, these methods often fail to simultaneously preserve local cluster structure and global data geometry. This article explores the application of power-weighted path metrics for the analysis of single cell RNA data.Extensive experiments on single cell RNA sequencing data sets confirm the usefulness of path metrics for dimension reduction and clustering. Distances between cells are measured in a data-driven way which is both density sensitive (decreasing distances across high density regions) and respects the underlying data geometry. By combining path metrics with multidimensional scaling, a low dimensional embedding of the data is obtained which respects both the global geometry of the data and preserves cluster structure. We evaluate the method both for clustering quality and geometric fidelity, and it outperforms other algorithms on a wide range of bench marking data sets.
Publisher
Cold Spring Harbor Laboratory