Related Experiment Video
Updated: Jun 26, 2026
08:35
Identification of Alternative Splicing and Polyadenylation in RNA-seq Data
Published on: June 24, 2021
scDeepAPA: a deep learning framework for single-cell alternative polyadenylation identification
Jialu Liang1, Qing Wang1, Sen Guo1
1Department of Health Outcomes and Biomedical Informatics, University of Florida, 1889 Museum Rd, Suite 7000, Gainesville, FL 32611, United States.
Briefings in Bioinformatics
|June 24, 2026
Summary
scDeepAPA is a new deep learning tool for analyzing alternative polyadenylation (APA) in single-cell RNA sequencing data. It accurately identifies polyadenylation sites and quantifies APA events, revealing cell-type-specific changes in diseases like Alzheimer's and cancer.
Area of Science:
- Genomics
- Computational Biology
- Molecular Biology
Background:
- Alternative polyadenylation (APA) is a key post-transcriptional regulator impacting transcript diversity and function.
- Single-cell RNA sequencing (scRNA-seq) offers potential for studying cell-type-specific APA dynamics.
- Existing computational tools for APA are often ill-suited for scRNA-seq data due to limitations with gene annotations and bulk data focus.
Purpose of the Study:
- To develop a deep learning framework, scDeepAPA, optimized for accurate polyadenylation site (PAS) detection and APA event quantification in scRNA-seq data.
- To enable functional interpretation of APA dynamics at single-cell resolution.
- To overcome limitations of existing tools in single-cell transcriptomic analyses.
Main Methods:
- scDeepAPA integrates convolutional feature extraction with Mamba-based state-space modeling and bidirectional LSTM layers.
- The framework is trained on high-confidence annotations from PolyASite v3.0 for robust PAS prediction.
- Performance is benchmarked against state-of-the-art PAS prediction models using human and mouse datasets.
Main Results:
- scDeepAPA demonstrates superior performance in accuracy, F1 score, and ROC metrics compared to existing models.
- Application to Alzheimer's disease data reveals cell-type-specific APA remodeling in immune and glial cells, with shifts towards proximal PAS usage.
- Analysis of KRAS-mutant lung cancer data identifies global proximal PAS activation and tumor-specific intronic APA events generating potential neoantigens.
Conclusions:
- scDeepAPA provides an accurate and efficient method for PAS identification and APA profiling in single-cell transcriptomics.
- The tool facilitates in-depth analysis of regulatory mechanisms and immunogenic consequences of APA.
- scDeepAPA advances the understanding of post-transcriptional regulation in neurodegeneration and cancer by enabling single-cell resolution studies.
