scDFC: A deep fusion clustering method for single-cell RNA-seq data

Abstract

Clustering methods have been widely used in single-cell RNA-seq data for investigating tumor heterogeneity. Since traditional clustering methods fail to capture the high-dimension methods, deep clustering methods have drawn increasing attention these years due to their promising strengths on the task. However, existing methods consider either the attribute information of each cell or the structure information between different cells. In other words, they cannot sufficiently make use of all of this information simultaneously. To this end, we propose a novel single-cell deep fusion clustering model, which contains two modules, i.e. an attributed feature clustering module and a structure-attention feature clustering module. More concretely, two elegantly designed autoencoders are built to handle both features regardless of their data types. Experiments have demonstrated the validity of the proposed approach, showing that it is efficient to fuse attributes, structure, and attention information on single-cell RNA-seq data. This work will be further beneficial for investigating cell subpopulations and tumor microenvironment. The Python implementation of our work is now freely available at https://github.com/DayuHuu/scDFC.

Unsupervised Gene-Cell Collective Representation Learning with Optimal Transport

“…To better learn the cell topology, scTAG (Yu et al 2022) adopts the topological adaptive graph convolutional networks (Du et al 2017) to extract the structural information of scRNA-seq data at different scales. scDFC (Hu et al 2023) applies the graph attention network (Veličković et al 2017) to reduce the noise effect brought in to the cell graph. As a novel approach, contrastive learning is introduced for cell clustering by scNAME (Wan, Chen, and Deng 2022) and scDCCA (Wang et al 2023a).…”

Mapping Cell Atlases at the Single‐Cell Level

“…Currently, nonsequence‐based models are capable of holistically accounting for all stages of scRNA‐seq analysis, [ 244 ] including normalization such as scVI, [ 245 ] data correction such as ResNet [ 246 ] and DESC, [ 247 ] clustering and cell annotation such as scVAE [ 248 ] and scDFC, [ 249 ] cell–cell communication analysis, [ 250 ] and RNA velocity such as DeepVelo. [ 251 ] Notably, Geneformer, [ 252 ] the first large model of computational biology, was pretrained on 30 million single‐cell transcriptomes to achieve predictions using transfer learning.…”

Effective multi-modal clustering method via skip aggregation network for parallel scRNA-seq and scATAC-seq data

“…Over the years, numerous attempts have been made to develop clustering methods for single-cell data [ 4 , 5 ]. Initially, the focal points of research revolved around fundamental clustering models such as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $k$\end{document} -means clustering and spectral clustering [ 6 , 7 ], along with their enhanced variants.…”