Title: Kernel density matching and its application for accurate alignment of single-cell RNA-seq samples
Authors: Mengjie Chen - University of Chicago (United States) [presenting]
Abstract: With technologies improved dramatically over recent years, single cell RNA-seq (scRNA-seq) has been transformative in studies of gene regulation, cellular differentiation, and cellular diversity. As the number of scRNA-seq datasets increases, a major challenge will be the standardization of measurements from multiple different scRNA-seq experiments enabling integrative and comparative analyses. However, scRNA-seq data can be confounded by severe batch effects and technical artifact. In addition, scRNA-seq experiments generally capture multiple cell-types with only partial overlaps across experiments making comparison and integration particularly challenging. To overcome these problems, we have developed a method, dmatch, which can both remove unwanted technical variation and assign the same cell(s) from one scRNA-seq dataset to their corresponding cell(s) in another dataset. By design, our approach can overcome compositional heterogeneity and partial overlap of cell types in scRNA-seq data. We further show that our method can align scRNA-seq data accurately across tissues biopsies.