T-Rex: Standardized Analysis of Germline Variants in Whole-Exome Sequencing Trios

Whole-exome sequencing (WES) enables the identification of rare germline variants contributing to pediatric diseases. Trio-based sequencing, comparing affected children with their parents, is particularly effective for rare disease genetics. However, WES data analysis requires bioinformatics expertise, varies across institutions, and is often incompatible with clinical workflows. We developed T-Rex (Trio Rare variant analysis of EXomes), a cross-platform desktop application that enables the standardized and local analysis of WES germline Trio data without the need for programming knowledge. T-Rex integrates state-of-the-art tools for alignment, dual-variant calling (GATK HaplotypeCaller + VarScan2), annotation (SNPEff/SNPSift), rare-variant filtering based on population frequencies (gnomAD), and family-based statistical testing, including the Transmission Disequilibrium Test with multiple-testing correction. Benchmarking of the dual-caller strategy on the Genome in a Bottle Ashkenazim Trio demonstrates high precision (99.2%) while maintaining robust sensitivity (91.1%). User testing (n=13) confirmed quick learning across clinicians and researchers. Application to a cohort of n=121 pediatric cancer Trio datasets, filtering for rare protein-coding variants (MAF[≤]0.1% in gnomAD v4.0), validated all assessable previously reported pathogenic variants. Overall, T-Rex enables clinicians to robustly analyze WES Trio data in compliance with data protection regulations without requiring additional software licenses. As one of the first platforms for comprehensive WES Trio analysis that requires no programming expertise while providing clinical-grade, end-to-end workflows, T-Rex facilitates collaborative research between clinics and reduces reliance on external providers. Implementation and AvailabilityThe source code is available on GitHub (https://github.com/SaraLuisaReh/trex). The fully precompiled app is available on Zenodo (https://zenodo.org/records/19135262).