Reproducible Bioinformatics Analysis Workflows for Detecting IGH Gene Fusions in B-Cell Acute Lymphoblastic Leukaemia Patients

Abstract

Simple Summary: B-cell acute lymphoblastic leukaemia (B-ALL) is a haematological malignancy driven by diverse genomic alterations, the most common being gene fusions, which can be detected via transcriptomic analysis. However, detecting gene fusions involving the Immunoglobulin Heavy Chain (IGH) region can be challenging due to its hyper variability. Our aim was to develop a workflow containing the algorithms FusionCatcher, Arriba, and STAR-Fusion, to achieve best-case sensitivity for IGH gene fusion detection. We analysed 35 patient samples harbouring IGH gene fusions and assessed the detection rates for each caller, before optimising the parameters to enhance sensitivity for IGH fusions. FusionCatcher and Arriba outperformed STAR-Fusion; however, by adjusting specific filtering parameters, we were able to improve STAR-Fusion's performance. This analysis highlights the importance of filtering optimization for IGH gene fusion events, offering alternative workflows for difficult-to-detect high-risk B-ALL alterations. B-cell acute lymphoblastic leukaemia (B-ALL) is characterised by diverse genomic alterations, the most frequent being gene fusions detected via transcriptomic analysis (mRNA-seq). Due to its hypervariable nature, gene fusions involving the Immunoglobulin Heavy Chain (IGH) locus can be difficult to detect with standard gene fusion calling algorithms and significant computational resources and analysis times are required. We aimed to optimize a gene fusion calling workflow to achieve best-case sensitivity for IGH gene fusion detection. Using Nextflow, we developed a simplified workflow containing the algorithms FusionCatcher, Arriba, and STAR-Fusion. We analysed samples from 35 patients harbouring IGH fusions (IGH::CRLF2 n = 17, IGH::DUX4 n = 15, IGH::EPOR n = 3) and assessed the detection rates for each caller, before optimizing the parameters to enhance sensitivity for IGH fusions. Initial results showed that FusionCatcher and Arriba outperformed STAR-Fusion (85–89% vs. 29% of IGH fusions reported). We found that extensive filtering in STAR-Fusion hindered IGH reporting. By adjusting specific filtering steps (e.g., read support, fusion fragments per million total reads), we achieved a 94% reporting rate for IGH fusions with STAR-Fusion. This analysis highlights the importance of filtering optimization for IGH gene fusion events, offering alternative workflows for difficult-to-detect high-risk B-ALL subtypes.