Activity Number:
|
179
- Statistical Methods in Single-Cell Transcriptomics
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Section on Statistics in Genomics and Genetics
|
Abstract #313417
|
|
Title:
|
NISC: Accurate Clustering Through Neural Network-Imputation for Single-Cell RNA-Seq Data with High Sparsity
|
Author(s):
|
Xiang Zhang* and Siyang Cao and Meng Lu and Nicholas Lytal and Lingling An
|
Companies:
|
University of Arizona and Department of Electrical & Computer Engineering, University of Arizona and Interdisciplinary Program in Statistics and Data Science, University of Arizona and Interdisciplinary Program in Statistics and Data Science, University of Arizona and Interdisciplinary Program in Statistics and Data Science, University of Arizona
|
Keywords:
|
Single-cell;
High sparsity;
Dropout;
Imputation;
Neural Network;
Autoencoder
|
Abstract:
|
High dimensional single-cell RNA sequencing (scRNA-seq) data become available recently to describe gene expression of tens of thousands of genes over increasing volume of single cells. However, this technology raises numbers of missing values (referred “dropouts”) caused by failures in amplification during the reverse-transcription of sequencing or other sources. The scRNA-seq data can reach more than 90% sparsity. Most existing imputation methods cannot handle the data with extremely high sparsity. We propose NISC (neural network-based imputation for single-cell RNA-seq data), an imputation approach that uses autoencoder with a novel loss function to recognize and correct the dropouts in scRNA-seq data. A series of comprehensive experiments show that NISC outperforms the available imputation methods on highly sparse scRNA-seq data.
|
Authors who are presenting talks have a * after their name.