Deep learning tackles single-cell analysis—a survey of deep learning for scRNA-seq analysis

Abstract

Since its selection as the method of the year in 2013, single-cell technologies have become mature enough to provide answers to complex research questions. With the growth of single-cell profiling technologies, there has also been a significant increase in data collected from single-cell profilings, resulting in computational challenges to process these massive and complicated datasets. To address these challenges, deep learning (DL) is positioned as a competitive alternative for single-cell analyses besides the traditional machine learning approaches. Here, we survey a total of 25 DL algorithms and their applicability for a specific step in the single cell RNA-seq processing pipeline. Specifically, we establish a unified mathematical representation of variational autoencoder, autoencoder, generative adversarial network and supervised DL models, compare the training strategies and loss functions for these models, and relate the loss functions of these models to specific objectives of the data processing step. Such a presentation will allow readers to choose suitable algorithms for their particular objective at each step in the pipeline. We envision that this survey will serve as an important information portal for learning the application of DL for scRNA-seq analysis and inspire innovative uses of DL to address a broader range of new challenges in emerging multi-omics and spatial single-cell sequencing.

scCapsNet-mask: an updated version of scCapsNet with extended applicability in functional analysis related to scRNA-seq data

“…Although many deep learning models have been adopted in the analysis of scRNA-seq data, most of them are limited to upstream analysis, and only a few examples are used in downstream functional analysis [4]. In addition to interpretable cell type identi cation, scCapsNet-mask could extend its applicability in functional analysis due to its unique category determining process (Architecture of CapsNet and the margin loss for classi cation).…”

“…Single cell RNA sequencing (scRNA-seq) could measure gene expression levels in individual cells and require diverse computational tools to deal with different computational tasks in the processing pipeline of scRNA-seq data [1]. The deep learning model can handle complex data well [2,3], and has been adopted in a series of necessary steps in the processing pipeline of scRNA-seq data, such as normalization, dimension reduction, and cell type identi cation [4]. However, the deep learning method lacks interpretability, which is usually operated as a 'block box' [5].…”

“…There have been several attempts to combine biological background to increase the interpretability of deep learning methods [6][7][8]. At the same time, the application of deep learning model usually focuses on the upstream analysis of scRNAseq data, and there are few methods to extend its applicability in downstream analysis, such as functional analysis, which is very important to reveal the biological meaning hidden in the data [4].…”

scCapsNet-mask: an updated version of scCapsNet with extended applicability in functional analysis related to scRNA-seq data

“…Single cell RNA sequencing (scRNA-seq) measures gene expression levels in individual cells and requires diverse computational tools to deal with different computational tasks in the processing pipeline [ 1 – 3 ]. The deep learning model can handle complex data well [ 4 , 5 ], and has been adopted in a series of necessary steps in the processing pipeline of scRNA-seq data, such as normalization, dimension reduction, and cell type identification [ 6 ]. However, the deep learning method lacks interpretability, which is usually operated as a ‘block box’ [ 7 ].…”

Editorial: Single cell intelligence and tissue engineering

“…Since the first single-cell RNA-sequencing (scRNA-seq) publication in 2009 ( Tang et al, 2009 ), single-cell-based technologies have generated massive datasets, offering great opportunities to fully address biomedical problems as well as posing a challenge to computational analysis. At the same time, machine learning methods have been successfully used in processing many kinds of big data, including scRNA-seq data analysis ( Petegrosso et al, 2020 ; Flores et al, 2022 ).…”