ExIR enables prioritizing driver and biomarker genes from omics data in a reference free manner

High-throughput sequencing enables genome-wide interrogation of biological systems, yet prioritizing functionally relevant genes and proteins from these data remains a key challenge. Here, we present ExIR (experimental data-based integrative ranking), a data-driven framework that classifies and ranks features as drivers, biomarkers, or mediators based on their behavior within inferred association networks. ExIR operates directly on experimental data without relying on external annotations. Across 14 transcriptomic and proteomic datasets, ExIR showed consistently strong performance in feature prioritization relative to commonly used methods. Application to RNA-seq data from a zebrafish model of mucopolysaccharidosis IIIA identified candidate regulators associated with disease progression. These results indicate that ExIR provides a generalizable approach for extracting biologically meaningful features from high-dimensional datasets, supporting more efficient downstream experimental investigation and interpretation.