Abstract:
Horizon is an integrated, ultra-fast genomic intelligence platform designed to streamline and enhance precision oncology workflows. The software delivers end-to-end analytical capabilities spanning somatic and germline variant interpretation, tumor profiling, and molecular diagnostics embedding agentic AI. Horizon ingests standard VCF files to perform highly accurate somatic variant detection, leveraging advanced algorithms to distinguish true driver alterations from technical noise. For germline variants, Horizon incorporates multi-omic evidence, population allele frequencies, and functional predictions to infer pathogenicity with clinical-grade confidence. To support harmonized clinical reporting, Horizon implements the Association for Molecular Pathology (AMP) guidelines, automatically classifying variants based on therapeutic, diagnostic, and prognostic relevance. Beyond interpretation, Horizon enables relevant panel development by designing cancer-specific, customizable oncology panels informed by prevalence data, mutational landscapes, and evidence-based biomarkers. Horizon further dissects tumor evolution and heterogeneity by reconstructing clonal architecture using variant allele frequencies, copy-number profiles, and evolutionary models. The platform also computes Tumor Mutational Burden (TMB) with high resolution, adjusting for panel size, sequencing depth, and artifact suppression to ensure accurate immunotherapy-related assessments. Additionally, Horizon deciphers tumor mutation Single base substitutions (SBS) signatures, identifying underlying mutagenic processes such as mismatch repair deficiency, or microsatellite instability (MSI). Integration of agentic AI allows to bring further insight into the interpretation of somatic variants. Together, these capabilities position Horizon as a comprehensive computational engine for cancer genomics, enabling rapid, reproducible, and clinically actionable insights to support both translational research and diagnostic decision-making.
