LYNXgwas

Locus anaYsis and geNomic eXplorer

A local GWAS visualization and fine-mapping platform. Point it at your GWAS summary statistics and it identifies loci, recovers rsIDs, computes LD, runs a full fine-mapping suite, and gives you an interactive Manhattan/LD/gene-track viewer per locus — all in your browser, entirely on your own machine. No data ever leaves your computer, and no cloud account or upload step is involved.

Why LYNXgwas

Going from a raw GWAS summary-statistics file to a set of well-characterized, fine-mapped loci usually means stitching together several separate tools by hand — a clumping step, a liftover or rsID lookup, a manual LD calculation, one-off scripts for each fine-mapping method, and finally some plotting code to actually look at the result. LYNXgwas wraps that whole workflow into one local application with a UI: define a project once, and loci identification, LD, annotation, and fine-mapping all run through the same pipeline and land in the same interactive viewer — including comparing the same locus across multiple GWAS datasets, and cross-ancestry fine-mapping across projects at once.

Screenshots

Home page — manage multiple GWAS projects, see status and locus/SNP counts at a glance:

Locus viewer — Manhattan plot, gene track, and LD triangle for a single locus:

Install

pip install lynxgwas

Requires a Java 11+ runtime on your machine (Adoptium is a good source if you don't have one).

Quick start

lynxgwas

This starts the local server and opens http://localhost:8765/ in your browser. On first run it will ask for (or offer to download) two things the pipeline needs:

• PLINK 1.9 — for LD computation and reference-panel subsetting (--plink <path> to skip the prompt)
• A GENCODE GFF3 annotation — for the gene track (--gff3 <path> to skip the prompt)

Both are remembered afterward. From the home page, use + New project to point LYNXgwas at your own GWAS summary statistics and walk through the setup wizard.

What it does

Loci & annotation

• Automatic loci identification from GWAS summary stats via PLINK clumping (or supply your own loci.txt)
• rsID recovery from a local dbSNP VCF, with optional NCBI/gnomAD API completion for anything left unmatched
• Gene-track annotation from a GENCODE GFF3, with nearest-gene lookup outside the plotted window
• Custom annotation tracks from your own TSV files (point/bar/flag styles) via an in-viewer wizard
• Genome-wide QC triage: genomic inflation factor (λ_GC), with an optional LDSC-intercept comparison to separate genuine polygenicity from confounding

LD & fine-mapping

• LD computation and pairwise LD triangles per locus (PLINK reference-panel subsetting under the hood)
• SuSiE — Bayesian sum-of-single-effects fine-mapping with credible sets
• FINEMAP-style Wakefield ABF fine-mapping for single-causal-variant loci
• GCTA-COJO — conditional & joint SNP selection with an automatic reliability/artifact check
• coloc — colocalization against a second trait's summary stats (PP.H0–H4)
• GWAMA — meta-analysis pass-through
• SuSiEx — cross-ancestry joint fine-mapping across multiple LYNXgwas projects at once
• Locus Matrix — compare the same locus region across many GWAS datasets side by side

Viewer & output

• Interactive Manhattan plot, gene track, and LD triangle, all zooming together
• Live locus editing: resize, split, create, and reorder loci without re-running the whole pipeline
• Per-locus and whole-genome PDF export; full Excel export of all annotated SNPs
• Multi-project management from a single home page, with per-project staleness tracking

Requirements

Documentation

Full architecture, Java package reference, API endpoints, and configuration keys: docs/ARCHITECTURE.md

License

MIT