The translation of complex cancer genomics into clinically actionable insights is a critical bottleneck in precision oncology. To automate the entire workflow, we present OncoGen.AI, an end-to-end, containerized platform that automates the process from raw sequencing data to a final clinical report. The platform integrates platform-agnostic pipelines (Illumina/Ion Torrent) for variant calling, copy number variation (CNV), and tumor mutational burden (TMB) analysis. Its core innovation is a clinically relevant Knowledge Graph comprising 20,242,772 triples across 11 node types, integrating data from domain-specific databases such as COSMIC, ClinVar, and others for evidence-based interpretation. The workflow culminates in a structured clinical report automatically generated by Google Gemini on the curated evidence extracted by the Knowledge Graph. When benchmarked against a diverse cohort of 13 clinical tumor exomes and validated across seven publicly available datasets, OncoGen.AI recovered 100% of pathogenic mutations identified by proprietary software (DRAGEN and Torrent Suite). Furthermore, in comparative analyses against other leading tools, OncoGen.AI demonstrated superior sensitivity, uniquely identifying clinically significant mutations in challenging Ion Torrent datasets that were otherwise missed. By providing a transparent "glass box" solution that bridges the gap between raw data and clinical action, OncoGen.AI is a comprehensive tool poised to accelerate reproducible decision-making in precision oncology.
Shome et al. (Fri,) studied this question.