B-cell acute lymphoblastic leukemia, or B-ALL, is a blood cancer that affects developing immune cells called B cells. Many cases are driven by fusion oncogenes, which form when pieces of different genes become incorrectly joined together. Detecting these fusions is important because they can help classify the leukemia, guide treatment decisions, and identify patients who may need more targeted care.

Researchers University of North Carolina at Chapel Hill have developed a new long-read fusion detection method called FUSILLI. The tool was designed specifically to improve detection of clinically relevant fusion transcripts in pediatric B-ALL.

Traditional short-read sequencing can sometimes miss or incompletely define fusion transcripts because the reads are too short to span complex RNA structures. Long-read whole-transcriptome sequencing, including nanopore sequencing, can capture longer RNA molecules, making it better suited for identifying full fusion transcripts and understanding how they are structured.

However, long-read sequencing data still needs the right computational tools. Many general-purpose fusion detection programs were not sensitive enough for precision diagnostics in B-ALL, especially when sequencing depth was limited. FUSILLI was developed to address this gap by focusing on leukemia-relevant fusions and evaluating how technical factors, such as read depth, affect detection.

The researchers tested FUSILLI using nanopore whole-transcriptome sequencing data from pediatric B-ALL samples. In the high-depth group, FUSILLI showed higher sensitivity than several existing fusion callers while maintaining strong specificity. This means it was better at finding true fusion events while still limiting false positives.

Sequencing depth was also an important part of the analysis. At lower read depth, all tools performed less well, but FUSILLI still outperformed the comparison methods. Computational down sampling suggested that around 10 million reads may be enough to detect important B-ALL fusions with good sensitivity using this approach.

This matters because affordable and accessible molecular diagnostics are especially important in pediatric cancer. A tool that works well with nanopore whole-transcriptome sequencing could help make fusion detection more practical in a wider range of clinical and research settings.

Overall, FUSILLI highlights how combining long-read RNA sequencing with disease-focused computational analysis can improve detection of cancer-driving fusion oncogenes. Better fusion detection may support more accurate diagnosis, improved leukemia subtyping, and more informed treatment planning for children with B-ALL.

Availability – https://github.com/jwanglab/fusilli/blob/main/fusilli.py