MultiGWAS-Explorer: an AI-assisted workflow for comparative analysis and visualization of stratified GWAS summary statistics

Large genome-wide association studies (GWASs) provide summary statistics across disorders, ancestries, and population strata, but comparing these datasets requires consistent allele handling, appropriate tests of between-group differences, and reproducible visualization. We developed MultiGWAS-Explorer, a locally executable workflow that harmonizes stratified GWAS summary statistics, compares effect estimates between strata, performs linkage-disequilibrium-aware locus prioritization, and generates genome-wide and locus-level figures. The statistical comparison is established; the main contribution is an integrated workflow that can be run from the command line or through an optional AI-assisted interface using the same versioned local scripts. Using sex- and ancestry-stratified Psychiatric Genomics Consortium schizophrenia GWAS summary statistics, quality control retained 18,906,333 eligible paired variants; differential lambda-GC values were 1.020, 1.021, and 0.997 for pooled, European, and Asian comparisons. On a 100,000-variant benchmark, MultiGWAS-Explorer showed no significance-threshold disagreement with two software GWAMA 2.2.2 or EasyStrata 8.6. In a 20-request Codex evaluation, 17/20 requests met all criteria initially and 19/20 did so after routing/error-handling improvements and clarification of three requests. Assessable configuration-generation runs that selected the intended task reproduced the command-line reference files exactly. No differential variant reached genome-wide significance. rs185665940 showed a suggestive, QC-sensitive signal ( P = 4.321 × 10 −7 ), whereas rs10166057 showed nominally differential evidence ( P = 4.5 × 10 −3 ). These findings are hypothesis-generating and do not establish genotype-by-sex interaction, conditional association, causal genes, or regulatory mechanisms.

Authors

Institutions

Publication Details

Journal
Frontiers in Genetics
Published
2026-11-19
DOI
https://doi.org/10.3389/fgene.2026.1922359
Primary Topic
Genetic Associations and Epidemiology
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

MultiGWAS-Explorer: an AI-assisted workflow for comparative analysis and visualization of stratified GWAS summary statistics

Zhong-shan Cheng
Frontiers in Genetics
Genetic Associations and Epidemiology
article

MultiGWAS-Explorer: an AI-assisted workflow for comparative analysis and visualization of stratified GWAS summary statistics

Zhong-shan Cheng
article en

Abstract

Large genome-wide association studies (GWASs) provide summary statistics across disorders, ancestries, and population strata, but comparing these datasets requires consistent allele handling, appropriate tests of between-group differences, and reproducible visualization. We developed MultiGWAS-Explorer, a locally executable workflow that harmonizes stratified GWAS summary statistics, compares effect estimates between strata, performs linkage-disequilibrium-aware locus prioritization, and generates genome-wide and locus-level figures. The statistical comparison is established; the main contribution is an integrated workflow that can be run from the command line or through an optional AI-assisted interface using the same versioned local scripts. Using sex- and ancestry-stratified Psychiatric Genomics Consortium schizophrenia GWAS summary statistics, quality control retained 18,906,333 eligible paired variants; differential lambda-GC values were 1.020, 1.021, and 0.997 for pooled, European, and Asian comparisons. On a 100,000-variant benchmark, MultiGWAS-Explorer showed no significance-threshold disagreement with two software GWAMA 2.2.2 or EasyStrata 8.6. In a 20-request Codex evaluation, 17/20 requests met all criteria initially and 19/20 did so after routing/error-handling improvements and clarification of three requests. Assessable configuration-generation runs that selected the intended task reproduced the command-line reference files exactly. No differential variant reached genome-wide significance. rs185665940 showed a suggestive, QC-sensitive signal ( P = 4.321 × 10 −7 ), whereas rs10166057 showed nominally differential evidence ( P = 4.5 × 10 −3 ). These findings are hypothesis-generating and do not establish genotype-by-sex interaction, conditional association, causal genes, or regulatory mechanisms.

Frontiers in GeneticsVol. 17
St. Jude Children's Research Hospital (US)
Genetic Associations and Epidemiology
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon