Single-cell RNA-seq denoising using a deep count autoencoder

Abstract

Single-cell RNA sequencing (scRNA-seq) has enabled researchers to study gene expression at a cellular resolution. However, noise due to amplification and dropout may obstruct analyses, so scalable denoising methods for increasingly large but sparse scRNA-seq data are needed. We propose a deep count autoencoder network (DCA) to denoise scRNA-seq datasets. DCA takes the count distribution, overdispersion and sparsity of the data into account using a negative binomial noise model with or without zero-inflation, and nonlinear gene-gene dependencies are captured. Our method scales linearly with the number of cells and can, therefore, be applied to datasets of millions of cells. We demonstrate that DCA denoising improves a diverse set of typical scRNA-seq data analyses using simulated and real datasets. DCA outperforms existing methods for data imputation in quality and speed, enhancing biological discovery.

Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression

“…We demonstrate that a regularization step, a commmon step in bulk RNA-seq analysis [Robinson et al, 2010, McCarthy et al, 2012 where parameter estimates are pooled across genes with similar mean abundance, can effectively overcome this challenge and yield reproducible models. Importantly, many statistical and deep-learning methods designed for single cell RNA-seq data utilize a negative binomial (or zero-inflated negative binomial) error model [Lopez et al, 2018, Eraslan et al, 2019. Our results suggest that each of these methods could benefit by substituting a regularized model, and that including an additional parameter for zero-inflation could exacerbate the risk of overfitting.…”

“…For example, ZINB-WaVE by Risso et al [2018] models counts as ZINB in a special variant of factor analysis. scVI and DCA also use the ZINB noise model [Lopez et al, 2018, Eraslan et al, 2019, either for normalization and dimensionality reduction in Bayesian hierarchical models, or for a denoising autoencoder. These pioneering approaches extend beyond pre-processing and normalization, but rely on the accurate estimation of per-gene error models.…”

A quantitative framework for evaluating single-cell data structure preservation by dimensionality reduction techniques

“…Systems biologists have derived and adapted numerical and computational methods for dimensionality reduction to allow for low-dimensional representation of single-cell data and deduction of cell states and fates (Van der Maaten and Hinton, 2008;Pierson and Yau, 2015;Linderman et al, 2017;Wang et al, 2017;Becht et al, 2018;Ding, Condon and Shah, 2018;Lopez et al, 2018;Risso et al, 2018;Eraslan et al, 2019;Townes et al, 2019).…”

“…However, given the distribution and sparsity of scRNA-seq data, complex, nonlinear transformations are often required to capture and visualize expression patterns. Unsupervised machine learning techniques and, more recently, deep learning methods, are being rapidly developed to assist researchers in single-cell transcriptomic analysis (Van der Maaten and Hinton, 2008;Pierson and Yau, 2015;Linderman et al, 2017;Wang et al, 2017;Becht et al, 2018;Ding, Condon and Shah, 2018;Lopez et al, 2018;Risso et al, 2018;Eraslan et al, 2019;Townes et al, 2019). Because these techniques condense cell features in the native space to a small number of latent dimensions for visualization, lost information can result in exaggerated or dampened cell-cell similarity.…”

Data-based RNA-seq Simulations by Binomial Thinning

“…The zero-inflated negative binomial distribution is sometimes used to model single-cell RNAseq data as it can account for the abundance of zeros observed in such data [Miao et al, 2018, Eraslan et al, 2019. A random variable y is distributed zero-inflated negative binomial, denoted y ∼ ZINB(π, µ, φ), if it is generated by the following hierarchical process:…”