scIAE: an integrative autoencoder-based ensemble classification framework for single-cell RNA-seq data

Abstract

Single-cell RNA sequencing (scRNA-seq) allows quantitative analysis of gene expression at the level of single cells, beneficial to study cell heterogeneity. The recognition of cell types facilitates the construction of cell atlas in complex tissues or organisms, which is the basis of almost all downstream scRNA-seq data analyses. Using disease-related scRNA-seq data to perform the prediction of disease status can facilitate the specific diagnosis and personalized treatment of disease. Since single-cell gene expression data are high-dimensional and sparse with dropouts, we propose scIAE, an integrative autoencoder-based ensemble classification framework, to firstly perform multiple random projections and apply integrative and devisable autoencoders (integrating stacked, denoising and sparse autoencoders) to obtain compressed representations. Then base classifiers are built on the lower-dimensional representations and the predictions from all base models are integrated. The comparison of scIAE and common feature extraction methods shows that scIAE is effective and robust, independent of the choice of dimension, which is beneficial to subsequent cell classification. By testing scIAE on different types of data and comparing it with existing general and single-cell–specific classification methods, it is proven that scIAE has a great classification power in cell type annotation intradataset, across batches, across platforms and across species, and also disease status prediction. The architecture of scIAE is flexible and devisable, and it is available at https://github.com/JGuan-lab/scIAE.

New Developments and Possibilities in Reanalysis and Reinterpretation of Whole Exome Sequencing Datasets for Unsolved Rare Diseases Using Machine Learning Approaches

“…However, single cell sequencing also requires better algorithms and computational power to analyze large datasets with much higher dimensions than non-single cell approaches. The use of deep learning approaches such as autoencoder algorithms has been shown to be quite effective in understanding important insights in cell biology [ 73 , 74 , 75 ]. There has already been an excellent scoping review here which covers the full scope of these approaches [ 76 ].…”

Clinical Phenotype Prediction From Single-cell RNA-seq Data using Attention-Based Neural Networks

“…Deep learning techniques have been applied to scRNA-seq analysis, showing promising results on many related tasks. For example, Yin et al (2022) propose an autoencoderbased classification framework to obtain compressed representations of scRNA-seq data. These representations are then fed into subsequent classifiers to predict the cell types.…”

Application of Deep Learning on Single-Cell RNA Sequencing Data Analysis: A Review

“…Using learned distance metric for label transfer and new cell discovery 2021 [140] sigGCN GAE/DFNN https://github.com/NabaviLab/sigGCN Concatenating latent representation learned from FFNN and GAE to predict cell type 2021 [141] scIAE AE https://github.com/JGuan-lab/scIAE Using ensemble of autoencoders with random projections to perform dimensionality reduction. Using the learned representations to train downstream classifiers for new data 2021 [142] mtSC DFNN https://github.com/bm2-lab/mtSC Using N-pair loss for deep metric learning across all reference datasets separately for trained model and using a consensus score from each reference dataset for cell annotation of query cell 2021 [143] ImmClassifier DFNN https://github.com/xliu-uth/ImmClassifier Using probability of coarse cell predictions into fine-grain predictions using the coarse grain probability distribution as input of a DFNN 2021 [144] netAE VAE https://github.com/LeoZDong/netAE Introduction of cell classification on latent representation for labeled cells and modularity loss based on cell–cell similarity matrix of latent representation 2021 [145] Cell BLAST VAE https://github.com/gao-lab/Cell_BLAST Using of improved distance-metric for mapping query cell to reference latent-representation and includes Poisson distribution as method for data augmentation of input scRNA-seq data 2020 [147] MultiCapsNet CapsNet [187] https://github.com/bojone/Capsule Using CapsNet for scRNA-seq data analysis 2021 [146] …”

“…However, such approaches are both labor- and resource-consuming. To address this, researchers are seeking deep learning approaches [126] , [127] , [128] , [129] , [130] , [131] , [132] , [133] , [134] , [135] , [136] , [137] , [138] , [139] , [140] , [141] , [142] , [143] , [144] , [145] , [146] that can handle this task with limited human supervision.…”