LocusComparePy

📦 R users: the original R package is locuscomparer.

News

• v0.1.0 (2026-07-08) — Initial Python port of locuscomparer, at feature parity with the R package v1.1.0: p-value / z-score (zscore_col) / logp_col input, custom LD via ld=, and the combined LocusCompare + LocusZoom figure via plotnine.

See NEWS.md for the full changelog.

1. Installation

LocusComparePy is a Python package for visualization of GWAS-eQTL colocalization events.

Install from PyPI (use the second line for the development version):

pip install locuscomparepy
# pip install git+https://github.com/boxiangliu/locuscomparepy

2. Example

To illustrate the use of locuscomparepy, we use the GWAS dataset from Nikpay et al. (2015) and the coronary artery eQTL dataset from GTEx v7 at the PHACTR1 locus:

import locuscompare as lc

gwas_fn = lc.example_path("gwas.tsv")
eqtl_fn = lc.example_path("eqtl.tsv")
fig = lc.locuscompare(in_fn1=gwas_fn, in_fn2=eqtl_fn,
                      title1="CAD GWAS", title2="Coronary Artery eQTL")
fig.save("locuscompare.png", width=10, height=5, dpi=150)

The output from locuscompare is a figure like the following:

The labeled SNP is the lead SNP (in this case for both studies), and other SNPs are colored according to their LD r² with the lead SNP.

3. Using your own dataset

The input to locuscompare() is a tab-delimited text file with two columns:

1. rsid
2. pval

Here is an example file:

rsid	pval
rs62156064	0.564395
rs7562234	0.399642
rs11677377	0.34308
rs35076156	0.625237

You can download the example files here: GWAS and eQTL datasets.

Then run the following commands:

import locuscompare as lc

gwas_fn = "path/to/gwas.tsv"
eqtl_fn = "path/to/eqtl.tsv"
fig = lc.locuscompare(in_fn1=gwas_fn, in_fn2=eqtl_fn, title1="GWAS", title2="eQTL")

You can also supply z-scores (zscore_col=) or pre-computed −log10(p) (logp_col=) instead of p-values, and bring your own LD via ld= (a SNP_A/SNP_B/R2 table) to bypass the reference database.

4. Documentation

To view documentation for each function, use help(function) (or function? in IPython/Jupyter). LocusComparePy exports the following functions:

Data munging

• assign_color: Assign color to each SNP according to LD.
• get_lead_snp: Get the lead SNP from the intersection of two studies.
• get_position: Append two columns, chromosome (chr) and position (pos), to the input.

Plotting

• locuscompare: Make a locuscompare plot.
• make_combined_plot: Combine two locuszoom plots with a locuscompare plot.
• make_locuszoom: Make a locuszoom plot.
• make_scatterplot: Make a scatter plot (the LocusCompare plot).

Data loading

• read_metal: Read association summary statistics from file.
• retrieve_LD: Retrieve SNP pairwise LD from the database.
• example_path: Path to a bundled example dataset (gwas.tsv, eqtl.tsv).

Position/LD lookups query a shared MySQL database over the network (non-standard port; some institutional firewalls may block it). Override the connection with LOCUSCOMPARE_DB_* environment variables.

5. Citation

If you use locuscompare, please cite the following paper: https://www.nature.com/articles/s41588-019-0404-0

Boxiang Liu, Michael J. Gloudemans, Abhiram S. Rao, Erik Ingelsson & Stephen B. Montgomery (2019) Abundant associations with gene expression complicate GWAS follow-up, Nature Genetics