Long-Read Whole-Transcriptome Sequencing and Selective Gene Panel Profiling Enable Sensitive Detection of Fusion...

Long-read whole-transcriptome sequencing (WTS) has the potential to precisely characterize fusion oncogenes that drive leukemia and other cancers. Although there is a variety of general-purpose fusion detection algorithms that use modern long-read sequencing data, they show poor sensitivity for precision diagnostics in B-cell acute lymphoblastic leukemia (B-ALL) and do not robustly assess technical and analytical parameters (eg, sequencing depth) to reliably detect fusion transcripts. FUSILLI (FUSions In Leukemia Long-read sequencing Investigator) is a novel long-read fusion detection algorithm, with a focus on targeted genomic subtyping in B-ALL. FUSILLI was evaluated against extant methods using nanopore WTS from 51 pediatric B-ALL samples sequenced at high depth and 68 at low depth (mean of 11.2 and 1.4 million reads, respectively). In the high-depth cohort, FUSILLI demonstrated increased sensitivity (0.81) compared with FusionSeeker, JAFFAL, and LongGF (0.63, 0.76, and 0.70, respectively), while maintaining high specificity (0.92). At lower sequencing depth, FUSILLI showed correspondingly lower sensitivity (0.27) but still outperformed the other fusion callers (sensitivities ranging from 0.09 to 0.16). Computational down sampling suggests that 10 million reads is sufficient to sensitively detect B-ALL-relevant fusions using this approach. FUSILLI detects B-ALL fusions with high sensitivity at modest sequencing depth, supporting implementation of nanopore WTS as a low-cost and globally accessible sequencing-based molecular diagnostic platform for pediatric B-ALL and other fusion-driven cancers.