NEBULA: A scalable and efficient framework for SNP-set analysis in the era of whole genome sequencing

Understanding the role of genetic variants in complex diseases is central to modern genomics. Genome-wide association and whole genome sequencing studies enable genotype–phenotype association discovery, but the growing scale of datasets poses computational challenges. Single nucleotide polymorphism set (SNP-set) analysis methods, which aggregate signals across groups of variants within genomic regions, pathways, or functional groups, enhance statistical power and help detect associations missed by single-variant analyses. Among these methods, our group proposed ABACUS, an algorithm based on a bivariate cumulative statistic that was proved to be robust to the presence of protective and deleterious SNP effects, as well as a mix of common and rare variants. However, its applicability to large-scale datasets is limited, mainly due to large memory usage and long execution time. In this study, we present NEBULA (Novel Entropy-Based framework for Unbiased Locus Analysis), a new methodology designed for SNP-set association analysis in case–control studies with binary phenotypes. NEBULA introduces a new computational architecture to address the limitations of the original method. Through three key innovations: (i) efficient memory management, (ii) parallelized execution, and (iii) optimized statistical computation, NEBULA significantly reduces computational burden and enables the analysis of large genomic datasets that were previously infeasible. Benchmarking results demonstrate that NEBULA enables the analysis of genomic datasets previously considered computationally infeasible. By improving performance while maintaining statistical rigor, NEBULA facilitates a deeper exploration of genetic contributions to complex traits.