ldblock package: linkage disequilibrium data structures

Vincent J. Carey, stvjc at channing.harvard.edu

Contents

1 Introduction

There is a nice vignette in snpStats concerning linkage disequilibrium (LD) analysis as supported by software in that package. The purpose of this package is to simplify handling of existing population-level data on LD for the purpose of flexibly defining LD blocks.

2 Import of HapMap LD data

The hmld function imports gzipped tabular data from hapmap’s repository .

suppressPackageStartupMessages({
 library(ldblock)
 library(GenomeInfoDb)
})
path = dir(system.file("hapmap", package="ldblock"), full=TRUE)
ceu17 = hmld(path, poptag="CEU", chrom="chr17")

## importing /tmp/RtmpkFe16b/Rinst52e41587e3623/ldblock/hapmap/ld_chr17_CEU.txt.gz

## done.

ceu17

## ldstruct for population CEU, chrom chr17
## dimensions: 36621 x 36621 ; statistic is Dprime 
## object structure:
##       ldmat       chrom      genome      allpos      poptag   statInUse 
## "dsCMatrix" "character" "character"   "numeric" "character" "character" 
##       allrs 
## "character"

3 A view of the block structure

For some reason knitr/render will not display this image nicely.

library(Matrix)

## 
## Attaching package: 'Matrix'

## The following object is masked from 'package:S4Vectors':
## 
##     expand

image(ceu17@ldmat[1:400,1:400], 
   col.reg=heat.colors(120), colorkey=TRUE, useRaster=TRUE)

This ignores physical distance and MAF. The bright stripes are probably due to SNP with low MAF.

4 Collecting SNPs exhibiting linkage to selected SNP

We’ll use ceu17 and the gwascat package to enumerate SNP that are in LD with GWAS hits.

library(gwascat)

## gwascat loaded.  Use makeCurrentGwascat() to extract current image.

##  from EBI.  The data folder of this package has some legacy extracts.

load(system.file("legacy/ebicat37.rda", package="gwascat"))
#seqlevelsStyle(ebicat37) = "NCBI"  # noop?
seqlevels(ebicat37) = gsub("chr", "", seqlevels(ebicat37))
e17 = ebicat37[ which(as.character(seqnames(ebicat37)) == "17") ]

Some dbSNP names for GWAS hits on chr17 are

rsh17 = unique(e17$SNPS)
head(rsh17)

## [1] "rs11078895" "rs11891"    "rs7501939"  "rs9905704"  "rs4796793" 
## [6] "rs78378222"

We will use expandSnpSet to obtain names for SNP that were found in HapMap CEU to have which \(D' > .9\) with any of these hits. These names are added to the input set.

length(rsh17)

## [1] 530

exset = ldblock::expandSnpSet( rsh17, ldstruct= ceu17, lb=.9 )

## Warning in ldblock::expandSnpSet(rsh17, ldstruct = ceu17, lb = 0.9): dropping
## 149 rsn not matched in ld matrix

length(exset)

## [1] 13209

all(rsh17 %in% exset)

## [1] TRUE

Not all GWAS SNP are in the HapMap LD resource. You can use your own LD data as long as the format agrees with that of the HapMap distribution.