medRxivGenetic & Genomic MedicinePreprint
LocusBlend: Flexible multi-index regional visualization of genomic association signals
Chenyu YangiD, Noah CookiD, Youjie ZengiD, Tianyi Fu, John BuddeiD, Carlos CruchagaiD, Michaël E. BelloyiD
Abstract
SummaryIt has become standard practice to visualize regional signals from genome-wide association studies (GWAS) using LocusZoom plots. Similarly, GWAS signals are compared to regionally matched quantitative trait loci (QTLs), i.e. variant-to-gene regulation data, using LocusCompare plots to aid assessment of candidate trait-related genes. Despite broad usage, these tools annotate variants by linkage disequilibrium (LD) to a single lead or index variant. This single-index representation has limitations for visualizing complex loci that contain multiple independent signals. We present LocusBlend, an interactive web application for multi-index LD-blended visualization of genomic loci. LocusBlend supports one or two genomic association summary-statistic datasets and one to three index variants, multi-index LocusZoom color-blended plots, and matching LocusCompare visualizations. Applications to Alzheimers disease GWAS and QTL signals illustrate LocusBlend enables visualization and separation of independent signals despite shared LD and high genomic complexity. Overall, LocusBlend is aimed at supporting researchers handle the continuously expanding complexity of human genomics findings. Availability and ImplementationLocusBlend is freely available at https://locusblend.wustl.edu. Publication ready plots are generated in <1min. Source code, documentation, example datasets, input templates, and reproducibility instructions are available at https://github.com/Belloy-Lab/LocusBlend. LocusBlend is implemented in Python using Streamlit, Plotly, and PLINK. Supplementary InformationSupplementary data are available at Bioinformatics online.
- organism: human
- assay: GWAS summary statistics
- research focus: software / tool
- research focus: GWAS / association mapping
- research focus: database / resource
- research focus: benchmarking
- approach: statistical model
- approach: benchmark study
In-depth summary
Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (21 Jul 2026) · CC BY 4.0
LocusBlend blends linkage disequilibrium colors for up to three index variants in LocusZoom and LocusCompare plots, helping separate independent signals at complex Alzheimer's disease loci.
Regional plotting tools such as LocusZoom and LocusCompare color variants by linkage disequilibrium (LD) to one lead variant, which obscures loci containing several independent GWAS or QTL signals, a situation made more common by larger studies and by TWAS/PWAS integration that detects sub-threshold signals. LocusBlend extends both plot types to two or three index variants using RGB-blended LD colors in a free interactive web tool.
Key results
ShownMulti-index LD blending gives each of two or three index variants its own color and mixes these colors by RGB rules, so variants tagging one signal versus several can be told apart within a single plot.
Each index is assigned a distinct color defined by a six-character hex code. Variants mostly in LD with one index keep that color, while those linked to multiple indexes get intermediate mixes. Diamonds mark index variants, and point size encodes maximal LD to any index. No quantitative evaluation is reported. · Overview; Index variant selection and LD calculation; Figure legends
SuggestedAt the female-biased Alzheimer's disease (AD) PSEN1 locus found by PWAS, LocusBlend plots guided by QTL lead variants split a region that was at first hard to interpret into 3 likely independent AD signals that are in LD with each other.
Female AD GWAS signals beneath the PSEN1 PWAS hit formed three nearby signals. They overlapped pQTLs for PSEN1 and PPAIP1 in brain, a PSEN1 eQTL in monocytes, a dorsolateral prefrontal cortex methylation QTL and a HEATR4 eQTL in blood. The separation was judged visually. · PSEN1; Figure legends
SuggestedAt the CD84 locus, multi-index plots indicated that only 2 of 4 independent female AD GWAS signals underlie the CD84 PWAS association. The other 2 reflect separate AD signals that colocalize with QTLs of other genes.
The 4 signals overlapped pQTLs for CD84 and FCER1G in cerebrospinal fluid, FCER1G eQTLs in microglia and monocytes, and an NDUFS2 eQTL in blood. The attribution rests on visual LD patterns, with no statistic reported for it. · Application to complex AD loci; CD84
ShownLocusBlend accepts one or two GWAS or QTL summary-statistic files. One file yields a locus zoom plot; two files add locus compare plots, drawn either as separate per-index panels or as one combined multi-index panel.
By default LD comes from PLINK calculations on an embedded 1000 Genomes reference, or users can supply precomputed LD. Choosing a single index gives a standard locus zoom plot; two or three indexes switch on multi-index mode. · Input data and analysis modes; Index variant selection and LD calculation; Figure legends
ClaimedThe tool is offered as a public web server that needs no registration and as publicly available Python source code that can be deployed locally, with documentation, example data and input templates for reproducing the manuscript plots.
The abstract states that publication-ready plots are produced in <1min but gives no hardware, locus size or variant count. The web app is free for non-commercial use and is built on Streamlit, Plotly and PLINK. · Abstract; Availability, reproducibility, and limitations; Data availability; Code availability
Methods, data and reproducibility
Streamlit/Plotly web app that takes one or two GWAS/QTL summary-statistic files, a locus and one to three index variants, obtains LD from PLINK on 1000 Genomes or user input, and draws locus zoom and locus compare plots with RGB-blended per-index colors and point size set by maximal LD; applied to two female-stratified Alzheimer's disease PWAS loci.
- Female-stratified Alzheimer's disease GWAS and brain-pQTL PWAS summary statistics at the PSEN1 and CD84 regions, from a prior sex-stratified study (Cook et al. 2025) · human (Alzheimer's disease GWAS)
- Molecular QTL summary statistics overlapping the loci: brain and cerebrospinal fluid pQTLs, monocyte, microglial and blood eQTLs, and a dorsolateral prefrontal cortex methylation QTL
- 1000 Genomes reference panel embedded in the app for default PLINK LD calculation · human (1000 Genomes Project panel)
- Example datasets and input templates for reproducing the manuscript's demonstration plots, hosted in the GitHub repository
- Methods & tools
- LocusBlend, Streamlit, Plotly, PLINK (LD calculation), 1000 Genomes reference panel, LocusZoom (plot type extended), LocusCompare (plot type extended)
- Data & accessions
- https://github.com/Belloy-Lab/LocusBlend
- Software & environment
- Python application built with Streamlit and Plotly, using PLINK and an embedded 1000 Genomes reference panel for LD; no package versions or container image are stated.
- Assay conditions & compute
- Publication-ready plots are reported to generate in <1min; no hardware or memory details are given.
- Reproducibility gaps
- The complete sex-stratified AD GWAS summary statistics must be obtained under the Cook et al. 2025 data availability terms, and the QTL data from the original studies. The web app is free only for non-commercial use, and no software versions are specified.
Limitations (3)
- LocusBlend only visualizes data. It does not perform statistical fine-mapping, conditional analysis or colocalization, so any candidate independent signals it reveals still need confirmation from complementary analyses. · stated by the authors
- Signal separation at PSEN1 and CD84 is judged by eye. This manuscript reports no conditional analysis or colocalization statistics for the separated signals, and no user study comparing interpretability against single-index plots. · noted in this summary
- Blending is capped at three index variants, so a locus such as CD84 with 4 reported signals cannot be fully shown in one plot. The manuscript also does not discuss how well the ancestry of the default 1000 Genomes LD panel matches user cohorts. · noted in this summary
Read this if you inspect GWAS, PWAS or QTL loci with several independent signals and want LocusZoom/LocusCompare-style plots that color variants by LD to up to three index variants.
- Citations
- 0
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- LocusZoom.js: Interactive and embeddable visualization of genetic association study results
Andrew P. Boughton, Ryan Welch, … Michael Boehnke·BioinformaticsPeer-reviewed·Mar 16, 2021
- 3D-Manhattan: An interactive visualization tool for multiple GWAS results
Shumpei Hashimoto·Ikushugaku zasshiPeer-reviewed·Jan 1, 2026
- ezQTL: A Web Platform for Interactive Visualization and Colocalization of Quantitative Trait Loci and GWAS
Tongwu Zhang, Alyssa Klein, … Kevin M. Brown·Genomics Proteomics & BioinformaticsPeer-reviewed·May 25, 2022
- Assocplots: a python package for static and interactive visualization of multiple-group GWAS results
Ekaterina A Khramtsova, Barbara E. Stranger·bioRxiv·Jul 7, 2016
- gwaRs: an R shiny web application for visualizing genome-wide association studies data
Lethukuthula L. Nkambule·bioRxiv·Apr 17, 2020
- flashfm-ivis: interactive visualisation for fine-mapping of multiple quantitative traits
Feng Zhou, Adam S. Butterworth, Jennifer L. Asimit·BioinformaticsPeer-reviewed·Jul 6, 2022