Autoencoder-based cluster ensembles for single-cell RNA-seq data analysis

Abstract

Background: Single-cell RNA-sequencing (scRNA-seq) is a transformative technology, allowing global transcriptomes of individual cells to be profiled with high accuracy. An essential task in scRNA-seq data analysis is the identification of cell types from complex samples or tissues profiled in an experiment. To this end, clustering has become a key computational technique for grouping cells based on their transcriptome profiles, enabling subsequent cell type identification from each cluster of cells. Due to the high feature-dimensionality of the transcriptome (i.e. the large number of measured genes in each cell) and because only a small fraction of genes are cell type-specific and therefore informative for generating cell type-specific clusters, clustering directly on the original feature/gene dimension may lead to uninformative clusters and hinder correct cell type identification.Results: Here, we propose an autoencoder-based cluster ensemble framework in which we first take random subspace projections from the data, then compress each random projection to a low-dimensional space using an autoencoder artificial neural network, and finally apply ensemble clustering across all encoded datasets for generating clusters of cells. We employ four evaluation metrics to benchmark clustering performance and our experiments demonstrate that the proposed autoencoder-based cluster ensemble can lead to substantially improved cell type-specific clusters when applied with both the standard k-means clustering algorithm and a state-of-the-art kernel-based clustering algorithm (SIMLR) designed specifically for scRNA-seq data. Compared to directly using these clustering algorithms on the original datasets, the performance improvement in some cases is up to 100%, depending on the evaluation metrics used. Conclusions:Our results suggest that the proposed framework can facilitate more accurate cell type identification as well as other downstream analyses. The code for creating the proposed autoencoder-based cluster ensemble framework is freely available from https://github.com/gedcom/autoencoder_cluster_ensemble

Predicting cell-to-cell communication networks using NATMI

“…At a detection rate threshold of 20% (commonly applied to single-cell datasets 11 , 25 ), most cell types in the Tabula Muris dataset expressed over a hundred ligands and receptors, with hematopoietic cell types expressing fewer ligands/receptors than other lineages (Supplementary Fig. 4a ).…”

Benchmarking clustering algorithms on estimating the number of cell types from single-cell RNA-sequencing data

“…An example in this fourth category is densi-tyCut [32], which estimates the number of cell types from a given dataset by modelling the density of cell distributions for generating a hierarchical cluster tree and subsequently selecting clusters that are most stable in the hierarchical cluster tree. In this study, we propose an alternative stability-based approach by taking advantage of scCCESS, a random sampling-based ensemble deep clustering model, previously proposed for scRNA-seq data clustering [33] for estimating the number of cell types. Our key assumption is that clustering from using the optimal number of clusters would be the most robust to small perturbations in the data, such as those introduced by random resampling, compared to those generated under the suboptimal number of clusters.…”

Grad Centroid Activation Mapping for Convolutional Neural Networks

“…Even if such layers are often used, this is not always the case. Several methods are still proposed that only use a simple autoencoder [18].…”