Eads, J. R. | bioRxiv Science

bioRxiv2025

Optical genome mapping identifies a novel PSIP1::TBL1X fusion in metastatic pancreatic neuroendocrine tumors

Background and aimsEffective treatment of metastatic neuroendocrine tumors (NETs) is limited by a lack of targeted therapies and clinically useful predictive biomarkers. We applied complementary genomic profiling technologies to identify genomic alterations that could lead to actionable drug targets and biomarkers. MethodsOptical genome mapping (OGM) and whole exome sequencing (WES) were performed on 70 liver metastases of neuroendocrine tumors from multiple anatomical primary sites. PacBio Iso-Seq long read sequencing and western blotting was used to confirm fusion gene expression. Findings were validated by rtPCR and western blotting on a separate cohort of 17 resection specimens from 14 pancreatic NET cases. ResultsSomatic variant calling from OGM data detected recurrent fusions involving TBL1X (PSIP1::TBL1X) and BEND2 (CHD7::BEND2 and NEO1::BEND2) in pancreatic neuroendocrine tumors (pNETs). BEND2 fusions were associated with high grade tumors, as previously described. The novel PSIP1::TBL1X fusion was confirmed to result in fusion transcript expression by both long-read sequencing and nested rtPCR. Presence of the PSIP1::TBL1X fusion transcript was also confirmed in a second cohort of pNET resection specimens and fusion protein expression was established by western blotting. Both TBL1X and BEND2 fusions are mutually exclusive with ATRX/DAXX mutations. ConclusionThis study underscores the importance of structural variants, including fusion genes as both prognostic biomarkers and potential therapeutic targets, especially in ATRX/DAXX negative pNETs. BEND2 fusion positivity may be a clinically useful biomarker for aggressive disease. As all TBL1X fusion isoforms contain the entire TBL1X coding sequence, TBL1X inhibition could be explored as a novel treatment for TBL1X fusion positive pNETs.

cancer biology↗

bioRxiv2018

Deep Semantic Protein Representation for Annotation, Discovery, and Engineering

Computational assignment of function to proteins with no known homologs is still an unsolved problem. We have created a novel, function-based approach to protein annotation and discovery called D-SPACE (Deep Semantic Protein Annotation Classification and Exploration), comprised of a multi-task, multi-label deep neural network trained on over 70 million proteins. Distinct from homology and motif-based methods, D-SPACE encodes proteins in high-dimensional representations (embeddings), allowing the accurate assignment of over 180,000 labels for 13 distinct tasks. The embedding representation enables fast searches for functionally related proteins, including homologs undetectable by traditional approaches. D-SPACE annotates all 109 million proteins in UniProt in under 35 hours on a single computer and searches the entirety of these in seconds. D-SPACE further quantifies the relative functional effect of mutations, facilitating rapid in silico mutagenesis for protein engineering applications. D-SPACE incorporates protein annotation, search, and other exploratory efforts into a single cohesive model.

bioinformatics↗