scLTF: A self-training clustering framework for large-scale single-cell RNA-Seq data
- 1School of Artificial Intelligence, Changchun Normal University, Changchun, China
Abstract
Single-cell RNA sequencing (scRNA-seq) technology has rapidly advanced in recent years, driving significant breakthroughs in developmental biology, cancer research, immunology, and other related fields. However, existing clustering methods still face performance bottlenecks when handling large-scale scRNA-seq data. To address this challenge, we propose a self-training clustering method for large-scale single-cell RNA-seq data, termed scLTF (Single-cell Louvain-Transformer Framework for Large-scale Clustering). This method first generates preliminary clusters using a fast Louvain algorithm and then selects key cells based on a custom “Cluster Representativeness Index”. Subsequently, a Transformer model is employed for iterative representation learning and optimization to optimize the final clustering results. Experiments on multiple real-world datasets demonstrate that scLTF achieves overall superior or comparable performance in clustering accuracy compared with several state-of-the-art deep clustering methods, with evaluation metrics (ARI, NMI, and ACC) improving on average by approximately 3.9%–24.1%; meanwhile, its runtime is only about 6.7%–43.9% of existing deep learning-based methods. scLTF integrates the advantages of traditional graph-based clustering and deep learning models, achieving both high clustering performance and computational efficiency, thus providing a practical and reliable tool for biomedical research and cellular analysis.
Related
- BILBO: a self-hosted platform for large-scale RNA-Seq analysis with a functional annotation clustering framework.
- scDAEC: deep clustering of single-cell data based on dual autoencoder
- Resolving Heterogeneous Mechanical Domains via Physics-Aware Deep Clustering of Single-Molecule Force Spectroscopy Data
- Estimating the completeness of large-scale single-cell sequencing projects
- Characterization of Single Ribosomes and Virus Particles with Large-Diameter Perfringolysin O Nanopores
BibTeX
@article{yuan2026scltf,
title = {{scLTF: A self-training clustering framework for large-scale single-cell RNA-Seq data}},
author = {Hongyi Yuan and Chunyan Wang and Qiucheng Sun and Mingyu Sun and Mujie Fan},
journal = {PLOS One},
year = {2026},
month = sep,
doi = {10.1371/journal.pone.0359153},
url = {https://doi.org/10.1371/journal.pone.0359153},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - scLTF: A self-training clustering framework for large-scale single-cell RNA-Seq data
AU - Hongyi Yuan
AU - Chunyan Wang
AU - Qiucheng Sun
AU - Mingyu Sun
AU - Mujie Fan
PY - 2026
DA - 2026/09/24
JO - PLOS One
DO - 10.1371/journal.pone.0359153
UR - https://doi.org/10.1371/journal.pone.0359153
AB - Single-cell RNA sequencing (scRNA-seq) technology has rapidly advanced in recent years, driving significant breakthroughs in developmental biology, cancer research, immunology, and other related fields. However, existing clustering methods still face performance bottlenecks when handling large-scale scRNA-seq data. To address this challenge, we propose a self-training clustering method for large-scale single-cell RNA-seq data, termed scLTF (Single-cell Louvain-Transformer Framework for Large-scale Clustering). This method first generates preliminary clusters using a fast Louvain algorithm and then selects key cells based on a custom “Cluster Representativeness Index”. Subsequently, a Transformer model is employed for iterative representation learning and optimization to optimize the final clustering results. Experiments on multiple real-world datasets demonstrate that scLTF achieves overall superior or comparable performance in clustering accuracy compared with several state-of-the-art deep clustering methods, with evaluation metrics (ARI, NMI, and ACC) improving on average by approximately 3.9%–24.1%; meanwhile, its runtime is only about 6.7%–43.9% of existing deep learning-based methods. scLTF integrates the advantages of traditional graph-based clustering and deep learning models, achieving both high clustering performance and computational efficiency, thus providing a practical and reliable tool for biomedical research and cellular analysis.
SN - 1932-6203
ER -