Software

GNExT (GWAS Network Exploration Tool)

GNExT is an integrated framework for building web-based platforms that visualize and interpret user-defined collections of genome-wide association study (GWAS) datasets, offering variant-, gene- and trait-centric exploration. A configurable Nextflow pipeline automates data preprocessing, so researchers can deploy their own instances without depending on curation by external platform maintainers. Using MAGMA, the pipeline aggregates SNP-level summary statistics into gene-level P values while accounting for local linkage disequilibrium, which increases statistical power compared with single-variant analyses. Through its integration with the network medicine tool Drugst.One, GNExT uses prioritized genes as seeds to identify disease modules and drug repurposing candidates, generating mechanistic hypotheses from GWAS summary statistics alone. Its scalability has been demonstrated on the Pan-UK Biobank European-ancestry dataset, which comprises 7,160 traits and more than 2 TB of data.

The platform source code is available at https://github.com/DyHealthNet/gnext_platform, the preprocessing pipeline at https://github.com/DyHealthNet/gnext_nf_pipeline, and the manuscript at https://doi.org/10.1038/s41588-026-02708-6 (Nature Genetics, 2026).

NApy

NApy is a Python package for efficient statistical testing on large-scale, mixed-type datasets with missing values, such as those from population-based cohort studies. It implements seven commonly used statistical tests (Pearson and Spearman correlation, χ²-test, t-test, Mann–Whitney U, ANOVA and Kruskal–Wallis) that cover all combinations of continuous, dichotomous and categorical variables, and it reports test statistics, effect sizes and P values with optional multiple-testingcorrection. Missing values are removed pairwise during computation, which avoids both the information loss of complete-case analysis and the potential bias of imputation. The tests are implemented in C++ with OpenMP parallelization and in Numba, which allows shared-memory multithreading without duplicating data across processes. In benchmarks on simulated data and the CHRIS cohort study, NApy outperformed established Python libraries and baseline implementations by up to several orders of magnitude in runtime, while keeping memory consumption low, enabling real-time analyses in interactive applications.

The software is available at https://github.com/DyHealthNet/NApy, and the manuscript at https://doi.org/10.1093/gigascience/giaf140 (GigaScience, 2025).