Poster #6 - Karleena Rybacki

Long-Read Whole-Transcriptome Sequencing Uncovers Gene Fusions and Structural Rearrangements in Fusion Panel-Negative Gliomas

Karleena Rybacki (MS) 1,2 Feng Xu (BS) 2,3 Joe Chan (BS) 2 Hannah Deutsch (BS) 4 Eloise Gaudet (BSE) 1, 5 Mian Umair Ahsan (MS) 2,6 Zizhuo Liang (BS) 2 Yuanquan Song (Ph.D) 2,4,7 Marilyn Li (MD, MS) 2,3,7 Kai Wang (Ph.D) 1,2,7 1. Department of Bioengineering, University of Pennsylvania,

Gene fusion (GF) detection in cancer diagnosis labs primarily relies on short-read sequencing fusion panels, such as the CHOP Fusion Panel which targets 119 oncogenes. These panels are designed to detect a predefined set of recurrent, well-characterized GFs. However, they have inherent limitations as short reads cannot span full-length transcripts, face alignment challenges in repetitive or complex genomic regions, and overlook GFs outside the list of targeted genes. To overcome these challenges and achieve a more comprehensive view of the transcriptomic and structural variant landscape, we performed Oxford Nanopore Technologies (ONT) long-read whole-transcriptome sequencing on ~50 high- and low-grade glioma samples previously deemed fusion panel-negative. We developed a computational analysis pipeline integrating GF detection, isoform detection and quantification, structural variant analysis, and fusion annotation/prioritization. From this, we identified and experimentally validated 20 unique GFs, including DUSP22::APOE, TCF12::CALM2, and CDKN2A::PTPRD which are not reported in established fusion databases (Mitelman, COSMIC Fusion, and ChimerDB 4.0). In addition, isoform-level analysis revealed alternative splicing patterns and structural variant analysis provided further characterization of genomic alterations. Together, these findings highlight the added value of ONT long-read whole-transcriptome sequencing for fusion panel-negative samples, resolving full-length isoforms, and capturing transcriptomic complexity in cancer.