Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Single-cell Curriculum Learning-based Deep Graph Embedding Clustering

Huifa Li, Jie Fu, Xinpeng Ling, Zhiyu Sun, Kuncan Wang, Zhili Chen

TL;DR

A single-cell curriculum learning-based deep graph embedding clustering (scCLG) that combines three optimization objectives, including topology reconstruction loss of cell graphs, zero-inflated negative binomial (ZINB) loss, and clustering loss, to learn cell-cell topology representation is proposed.

Abstract

The swift advancement of single-cell RNA sequencing (scRNA-seq) technologies enables the investigation of cellular-level tissue heterogeneity. Cell annotation significantly contributes to the extensive downstream analysis of scRNA-seq data. However, The analysis of scRNA-seq for biological inference presents challenges owing to its intricate and indeterminate data distribution, characterized by a substantial volume and a high frequency of dropout events. Furthermore, the quality of training samples varies greatly, and the performance of the popular scRNA-seq data clustering solution GNN could be harmed by two types of low-quality training nodes: 1) nodes on the boundary; 2) nodes that contribute little additional information to the graph. To address these problems, we propose a single-cell curriculum learning-based deep graph embedding clustering (scCLG). We first propose a Chebyshev graph convolutional autoencoder with multi-criteria (ChebAE) that combines three optimization objectives, including topology reconstruction loss of cell graphs, zero-inflated negative binomial (ZINB) loss, and clustering loss, to learn cell-cell topology representation. Meanwhile, we employ a selective training strategy to train GNN based on the features and entropy of nodes and prune the difficult nodes based on the difficulty scores to keep the high-quality graph. Empirical results on a variety of gene expression datasets show that our model outperforms state-of-the-art methods. The code of scCLG will be made publicly available at https://github.com/LFD-byte/scCLG.

Single-cell Curriculum Learning-based Deep Graph Embedding Clustering

TL;DR

A single-cell curriculum learning-based deep graph embedding clustering (scCLG) that combines three optimization objectives, including topology reconstruction loss of cell graphs, zero-inflated negative binomial (ZINB) loss, and clustering loss, to learn cell-cell topology representation is proposed.

Abstract

The swift advancement of single-cell RNA sequencing (scRNA-seq) technologies enables the investigation of cellular-level tissue heterogeneity. Cell annotation significantly contributes to the extensive downstream analysis of scRNA-seq data. However, The analysis of scRNA-seq for biological inference presents challenges owing to its intricate and indeterminate data distribution, characterized by a substantial volume and a high frequency of dropout events. Furthermore, the quality of training samples varies greatly, and the performance of the popular scRNA-seq data clustering solution GNN could be harmed by two types of low-quality training nodes: 1) nodes on the boundary; 2) nodes that contribute little additional information to the graph. To address these problems, we propose a single-cell curriculum learning-based deep graph embedding clustering (scCLG). We first propose a Chebyshev graph convolutional autoencoder with multi-criteria (ChebAE) that combines three optimization objectives, including topology reconstruction loss of cell graphs, zero-inflated negative binomial (ZINB) loss, and clustering loss, to learn cell-cell topology representation. Meanwhile, we employ a selective training strategy to train GNN based on the features and entropy of nodes and prune the difficult nodes based on the difficulty scores to keep the high-quality graph. Empirical results on a variety of gene expression datasets show that our model outperforms state-of-the-art methods. The code of scCLG will be made publicly available at https://github.com/LFD-byte/scCLG.
Paper Structure (23 sections, 16 equations, 4 figures, 3 tables, 1 algorithm)

This paper contains 23 sections, 16 equations, 4 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. PRELIMINARIES
  4. Notations
  5. Chebyshev Graph Convolution
  6. Proposed Approach
  7. Multi-Criteria ChebConv Graph Autoencoder
  8. Reconstruction Loss
  9. ZINB Loss
  10. Clustering Loss
  11. Curriculum Learning with Data Pruning
  12. Hierarchical Difficulty Measurer
  13. Data Pruning
  14. The Proposed scCLG Algorithm
  15. Experiments
  16. ...and 8 more sections

Figures (4)

  • Figure 1: Framework of scCLG. (A) Pre-training: pretraining the proposed ChebAE with adjacency matrix decoder and ZINB decoder. Then calculate node difficulty using a hierarchical difficulty measurer and prune the data. (B) Formal training: using all three criterias to optimize the model in more detail from easy to hard pattern with pruned data.
  • Figure 2: The model architecture of multi-criteria ChebAE. ChebAE integrates three loss components: reconstruction loss, ZINB loss, and a clustering loss to optimize the low-dimensional latent representation.
  • Figure 3: Parameter analysis. (A) Comparison of the average ARI and NMI values with different neighbor parameters $k$. (B) Comparison of the average ARI and NMI values with different numbers of genes.
  • Figure 4: Comparison of the average ARI and NMI values with different data pruning rates and pruning strategies.