Make Locus Zoom-like plots with your own LD matrix

This script creates an R function to create regional Manhattan plots with points coloured according to LD and genes annotated beneath.

Input file format

You will need at least two files to generate a Locus Zoom (LZ) plot: GWAS summary stats file, and a “gene” file.

The LD file can be generated on the fly, as long as you have access to the 1000GP reference data and the correct tools to calculate the LD (see Before you begin).

Please also note that the column names of the required files must match exactly to those specified below, so please take your time to check your headers and edit it to match with the specification.

GWAS summary statistics

Though you may have more information in your GWAS summary statistics, you only need the following four columns:

Instead of using P as significance, you may use logBF (e.g. from a trans-ancestry meta-analysis that used MANTRA).

Gene file

A file of the genes within the region for use in the annotation step.

This file must have four columns:

The UCSC_GRCh37_Genes_UniqueList.txt file is provided with the code.

LD file

This is a file that contains the LD between the SNP to be labelled (top-hit / SNP of interest) and the SNPs included in the GWAS summary statistics.

This file must contain the following two columns:

The SNP names of the LD file must match the SNP names in the GWAS summary statistics file.
Therefore, we recommend you to use rsID wherever possible to match with the 1000GP reference data (used to calculate the LD).

To calculate the LD yourself, follow the code below:

1
2

bcftools view -r CHR:START-END /path/to/reference/chr/vcf.gz -Oz -o /path/to/output/tmp.vcf.gz
plink1.9 --vcf /path/to/output/tmp.vcf.gz --allow-no-sex --snps-only --r2 inter-chr --ld-snp <rsid-of-interest> --ld-window-r2 0 --out /path/to/output/ld_file

where:

For more information on why this could be useful, take a look at Generate multiple LZ plots.

Examples

Simple Locus Zoom

You will be able to run the following code and produce the same Locus Zoom plot using the example data set from the GitHub repo.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

# Load necessary files into R
Example.assoc.linear <- read.delim("Example.assoc.linear", stringsAsFactors = FALSE, header = TRUE)
Example.ld <- read.table("Example.ld", stringsAsFactors = FALSE, header = TRUE)
UCSC_GRCh37_Genes_UniqueList.txt <- read.delim("Example.genes", stringsAsFactors = FALSE, header = TRUE)

# Load the locuszoom function into R
source("functions/locus_zoom.R")

# Create a LocusZoom-like plot
locus.zoom(data = Example.assoc.linear,
           region = c(16, 53340000, 54550000),
           offset_bp = 0,
           ld.file = Example.ld,
           genes.data = UCSC_GRCh37_Genes_UniqueList.txt,
           plot.title = "Association of FTO with BMI in Europeans",
           file.name = "Example.jpg",
           secondary.snp = c("rs1121980", "rs8060235"),
           secondary.label = TRUE)

Locus Zoom with coloured genes based on MAGMA output

This is not reproducible from the example data.

1
2
3
4
5
6
7
8

locus.zoom(data = EUR_meta_full1_clean_rsid.nfiltered_chr7,
           gene = "MLXIPL",
           offset_bp = 500000,
           genes.data = UCSC_GRCh37_Genes_UniqueList,
           plot.title = "Association of MLXIPL with gout in Europeans",
           file.name = "alternateExample.jpg",
           genes.pvalue = European_GWAS_2020_test.genes_named,
           colour.genes = TRUE)