The Hidden Sources of Spurious Fusion Transcripts in Plants

Overview

This repository contains the computational workflows, scripts, test datasets, and example files used in the study:

The Hidden Sources of Spurious Fusion Transcripts in Plants

Fusion transcripts were initially discovered in cancer and are increasingly reported across many organisms, including plants. However, whether most reported plant fusion transcripts represent genuine biological events or technical/analytical artifacts has remained unclear. In this study, we systematically re-evaluated fusion transcript detection in plants using multiple sequencing platforms, including:

• PacBio Iso-Seq (IsoSeq)
• Nanopore direct RNA sequencing (dRNA-seq)
• Nanopore PCR-cDNA sequencing (cDNAseq)
• Short-read RNA-seq (RNAseq)

Our results show that many reported plant fusion transcripts arise from previously underappreciated hidden sources, including:

• Short homologous sequence (SHS)-mediated template switching
• Alignment artifacts
• Genome assembly errors
• Annotation inconsistencies
• Reference-genome bias

This repository provides workflows to identify, classify, and visualize fusion transcripts in plants.

---

Associated Manuscript

The Hidden Sources of Spurious Fusion Transcripts in Plants (under revision)

---

Key Contributions

Conceptual Advances

This work proposes an important distinction:

• Fusion detection ≠ True biological fusion events

Many detected fusion signals can arise during sequencing, reverse transcription, alignment, or annotation processes.

Methodological Advances

This repository includes workflows for:

• Long-read fusion detection using an optimized JAFFAL
• Short-read fusion calling using Arriba
• Breakpoint classification (SHS / Joined / Unknown / Misaligned)
• Fusion network analysis
• Hybrid rice allelic/non-allelic fusion screening
• IGV-ready fusion visualization
• Overlap with chromatin/RNA 3D interaction datasets
• Genome annotation comparison

---

Repository Structure

.
├── README.md
├── workflow_commands.md
│
├── test_data/
│   Example datasets for running workflows
│
├── examples/
│   Example outputs used in manuscript analyses
│
├── figures/
│   Example visualization figures
│
├── environment/
│   software environments and R sessionInfo
│
├── scripts/
│   Main analysis scripts organized by module
│   ├── 01_preprocessing/
│   ├── 02_fusion_calling/
│   ├── 03_fusion_and_breakpoints_visualization/
│   ├── 04_breakpoint_classification/
│   ├── 05_fusion_network/
│   ├── 06_fusions_calling_short_reads/
│   ├── 07_hybrid_fusions/
│   ├── 08_others/
│   └── Stats_and_plot.R

---

Workflow Summary

01_preprocessing

Preprocessing of:

• PacBio Iso-Seq raw subreads
• Nanopore direct RNA / cDNA reads
• Read correction and alignment

---

02_long-read_fusion_calling

Detection of putative fusion transcripts from long-read data using an optimized version of JAFFAL.

---

03_fusion_visualization

Generate IGV-compatible files for:

• Fusion transcripts
• Fusion breakpoints

---

04_breakpoint_classification

Classify fusion breakpoints into four categories:

• SHS (short homologous sequence)
• Joined
• Unknown
• Misaligned

---

05_fusion_network

Network analysis of recurrent fusion-partner genes using igraph.

---

06_fusion_calling_short_reads

Detect fusion transcripts from short-read RNA-seq using:

• STAR
• Arriba

---

07_hybrid_fusions

Identify candidate allelic and non-allelic fusion events in hybrid rice using the trans-mate pair strategy.

---

08_others

Additional analyses:

• Overlap with 3D genomic interaction data
• MH63 HiFi genome assembly
• Annotation-version comparisons

---

Test Data

Small example datasets are provided under:

test_data/

These allow users to test the pipeline structure and reproduce representative outputs.

For full-scale analyses, raw sequencing data used in the study are publicly available at NCBI: PRJNA1291274

---

Citation

If you use this repository, please cite:

The Hidden Sources of Spurious Fusion Transcripts in Plants

(doi to be added after publication)

---

Contact

For questions, suggestions, or bug reports, please open an Issue or contact the corresponding authors.

• Xi-Tong Zhu E-mail: z724@qq.com
• Yidan Ouyang E-mail: diana1983941@mail.hzau.edu.cn
• Ling-Ling Chen E-mail: llchen@gxu.edu.cn

---

Final Note

This repository is not only a fusion-calling workflow. It is also a framework for asking a more fundamental question:

When we detect a fusion transcript, how confident can we be that a genuine fusion event has actually occurred?

We hope this resource helps future studies avoid false positives and improve rigor in transcriptomics research.

License

This project is licensed under the MIT License. See the LICENSE file for details.