Deep Denoising Sparse Coding

Abstract

Single-cell Ribonucleic Acid sequencing (scRNA-seq) has great potential to discover cell types, identify cell states, trace development lineages, and reconstruct the spatial organization of cells. Clustering transcriptomes profiled by scRNA-seq has been routinely conducted to reveal cell heterogeneity and diversity. In fact, scRNA-seq data contain an abundance of dropout events that lead to zero expression measurements. These dropout events may be the result of technical sampling effects or real biology arising from stochastic transcriptional activity. Therefore clustering analysis of scRNA-seq data remains a statistical and computational challenge. Here, we have developed Deep Denoising Sparse Coding (DDSC), a deep clustering method combine autoencoder and sparse coding approach. Based on six real datasets from five representative single-cell sequencing platforms, DDSC outperformed some state-of-the-art methods under various clustering performance metrics and exhibited improved scalability. Its accuracy and efficiency make DDSC a promising algorithm for clustering large-scale scRNA-seq data.