Gene fusion detection in long-read transcriptome sequencing data with GFvoter

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Gene fusions are common in cancer and serve as important diagnostic biomarkers and therapeutic targets.
• Long-read transcriptome sequencing offers advanced capabilities for detecting gene fusions.

Purpose of the Study:

• To develop GFvoter, a novel method for accurate gene fusion detection using long-read RNA sequencing data.
• To evaluate GFvoter's performance against existing methods in identifying gene fusions.

Main Methods:

• GFvoter utilizes a multivoting strategy, integrating outputs from two RNA-seq aligners and two fusion detection tools.
• A custom scoring mechanism is employed within the voting process to enhance fusion identification accuracy.
• The method was validated using simulated and real cell line datasets from PacBio and Nanopore sequencing platforms.

Main Results:

• GFvoter demonstrated significantly superior performance compared to alternative gene fusion detection methods.
• The tool successfully identified the RPS6KB1:VMP1 gene fusion in MCF-7 cell lines, a fusion missed by other tested tools.
• Validation on both simulated and real datasets confirmed GFvoter's high accuracy and reliability.

Conclusions:

• GFvoter accurately detects gene fusions from long-read RNA sequencing data.
• This method has the potential to enhance cancer diagnosis and guide therapeutic strategies.
• GFvoter is publicly available for research use.