Related Experiment Video
Updated: Jun 30, 2026
10:12
Droplet Barcoding-Based Single Cell Transcriptomics of Adult Mammalian Tissues
Published on: January 10, 2019
Transformers for single-cell RNA sequencing: a survey
1Department of Computer Science, New Jersey Institute of Technology, Guttenberg Information Technologies Center (GITC), Suite 4100, University Heights, Newark, New Jersey 07102, USA.
Briefings in Bioinformatics
|June 29, 2026
Summary
Transformers show promise for analyzing complex single-cell RNA sequencing (scRNA-seq) data, overcoming limitations of traditional methods. This survey details their application, performance, and future directions in biomedical research.
Area of Science:
- Computational biology
- Genomics
- Artificial intelligence in medicine
Background:
- Single-cell RNA sequencing (scRNA-seq) generates high-dimensional, sparse data challenging for conventional analysis.
- Transformers, a deep learning architecture, offer advanced capabilities for complex biological data.
- Existing methods struggle with scRNA-seq data characteristics like sparseness and batch effects.
Purpose of the Study:
- To provide a comprehensive overview of Transformer applications in scRNA-seq data analysis.
- To systematically analyze Transformer models for specific and multiple downstream tasks (foundation models).
- To examine Transformer models regarding performance, efficiency, interpretability, and scalability.
Main Methods:
- Systematic review and analysis of Transformer architectures applied to scRNA-seq.
- Categorization of Transformers into task-specific and foundation models.
- Evaluation of models based on performance, computational efficiency, interpretability, and scalability.
Main Results:
- Transformers significantly improve model performance in scRNA-seq analysis due to self-attention and transfer learning.
- Analysis covers both specialized Transformers and versatile foundation models.
- Identified key considerations for model selection and future research directions.
Conclusions:
- Transformers represent a powerful tool for advancing scRNA-seq data analysis.
- This survey offers a practical resource for researchers, addressing current challenges and guiding future development.
- Future research should explore Transformers beyond scRNA-seq and across other omics layers.
