A Dirichlet Mixture Model for Clustering Droplet-Based Single Cell Transcriptomic Data

• Activity Number: 408 - Methods for Single-Cell Genomic Analysis
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Wei Chen* and Zhe Sun and Ke Deng and Ming Hu
• Companies: University of Pittsburgh and University of Pittsburgh and Tsinghua University and Cleveland Clinic
• Keywords: single cell ; Dirichlet ; EM

Abstract

Single cell transcriptome sequencing (scRNA-seq) has become a revolutionary tool to study transcriptomic profiles of individual cells. The recently developed droplet-based platform enables efficient parallel processing of thousands of cells with direct counting of transcript copies using Unique Molecular Identifier (UMI). Despite the technology advance, statistical methods and computational tools are still lacking for analyzing droplet-based single cell transcriptomic data. We developed a Dirichlet mixture model for clustering droplet-based Single Cell transcriptomic data. This approach explicitly models UMI count data from scRNA-seq experiments and characterizes variations across different cell clusters via Dirichlet mixture prior. We performed comprehensive simulations and compared the performance with K-means and CellTree. In addition, we analyzed public and in-house scRNA datasets from multiple studies of complex diseases. We demonstrate that our method has a substantially improved clustering accuracy and stability and provides clustering uncertainty for downstream analysis and better biological interpretations. The program has been implemented in a user-friendly R package.