Anchored-fusion enables targeted fusion search in bulk and single-cell RNA sequencing data

Here, we present Anchored-fusion, a highly sensitive fusion gene detection tool. It anchors a gene of interest, which often involves driver fusion events, and recovers non-unique matches of short-read sequences that are typically filtered out by conventional algorithms. In addition, Anchored-fusion contains a module based on a deep learning hierarchical structure that incorporates self-distillation learning (hierarchical view learning and distillation [HVLD]), which effectively filters out false positive chimeric fragments generated during sequencing while maintaining true fusion genes. Anchored-fusion enables highly sensitive detection of fusion genes, thus allowing for application in cases with low sequencing depths. We benchmark Anchored-fusion under various conditions and found it outperformed other tools in detecting fusion events in simulated data, bulk RNA sequencing (bRNA-seq) data, and single-cell RNA sequencing (scRNA-seq) data. Our results demonstrate that Anchored-fusion can be a useful tool for fusion detection tasks in clinically relevant RNA-seq data and can be applied to investigate intratumor heterogeneity in scRNA-seq data. [Display omitted] •Anchored-fusion detects fusion genes with high sensitivity in paired-end RNA-seq•Anchoring a gene of interest avoids over-filtering based on homology alignment•A deep learning module filters false positive chimeric reads•Anchored-fusion shows high sensitivity in single-cell applications Gene fusion is one of the key events driving cancer development. Identifying critical fusion genes using RNA sequencing (RNA-seq) data has been applied in clinical samples for diagnosis, subtyping, and targeted therapeutic purposes. However, current gene fusion detection algorithms of RNA-seq are limited by their lack of sensitivity, making it difficult to apply them to low-read-depth data, for example in single-cell and/or clinical contexts. Yuan et al. present Anchored-fusion, a method for detecting fusion genes with high sensitivity from paired-end RNA-seq. Anchoring a gene of interest avoids over-filtering, and a deep learning model removes false positives. Anchored-fusion demonstrates superior sensitivity in various scenarios, particularly in detecting fusion genes from single-cell RNA-seq.