LDlinkR: An R Package for Rapidly Calculating Linkage Disequilibrium Statistics in Diverse Populations

Introduction

LDlink is an interactive and powerful suite of web-based tools for querying germline variants in human population groups of interest to generate interactive tables and plots. All population genotype data originates from Phase 3 (Version 5) of the 1000 Genomes Project and variant RS numbers are indexed based on dbSNP 155.

LDlinkR is an R package developed to query and download results (internet access required) generated by LDlink web-based applications from the R console. LDlinkR accelerates genomic research by providing efficient and user-friendly functions to programmatically interrogate pairwise linkage disequilibrium from large lists of genetic variants.

Please see the the sections below and the online LDlink documentation for more information about understanding linkage disequilibrium (LD) and additional details about how LDlink calculates patterns of LD across a variety of ancestral human populations.

Understanding Linkage Disequilibrium

What is linkage disequilibrium? Perhaps it is best to start with linkage equilibrium. Linkage equilibrium exists when alleles from two different genetic variants occur independently of each other. The inheritance of such variants follows probabilistic patterns governed by population allele frequencies. The vast majority of genetic variants on a chromosome are in linkage equilibrium. Variants in linkage equilibrium are not considered linked.

Linkage disequilibrium is present when alleles from two nearby genetic variants commonly occur together in a non-random, linked fashion. This linked mode of inheritance results from genetic variants in close proximity being less likely to be separated by a recombination event and thus alleles of the variants are more commonly inherited together than expected. Alleles of variants in linkage disequilibrium are correlated; with the degree of correlation generally greater in magnitude the closer the variants are in physical distance. Measures of linkage disequilibrium include D prime (D’) and R squared (R²).

A haplotype is a cluster of genetic variants that are inherited together. Humans are diploid; having maternal and paternal copies of each autosomal chromosome. Each chromosomal copy is organized into segments of high linkage disequilibrium, called haplotype “blocks”. Due to unique population histories and differences in variant allele frequencies, haplotype structure tends to be population specific. Although haplotypes are essential for calculating measures of linkage disequilibrium, haplotypes are seldom directly observed. Statistical chromosome phasing techniques are often necessary to infer individual haplotypes.

LDlink Data Sources

dbSNP (source: GRCh37 and GRCh38) - To investigate patterns of linkage disequilibrium, LDlink focuses on two main classes of genetic variation: single nucleotide polymorphisms (SNPs) and insertions/deletions (indels). Every module of LDlink requires the entry of at least one variant as identified by a RefSNP number (RS number) or genomic position (chr#:position). RS numbers are unique labels assigned by dbSNP and are well-curated identifiers that follow the format “rs” followed by a number. The current implementation of LDlink references dbSNP and only accepts input for bi-allelic variants.

1000 Genomes Project (source: GRCh37, GRCh38, and GRCh38 High Coverage) - Publicly available reference haplotypes from the 1000 Genomes Project are used by LDlink to calculate population-specific measures of linkage disequilibrium. Haplotypes are available for continental populations (ex: European, African, and Admixed American) and sub-populations (ex: Finnish, Gambian, and Peruvian). All LDlink modules require the selection of at least one 1000 Genomes Project sub-population, but several sub-populations can be selected simultaneously. Available haplotypes vary by sub-population based on sample size.

UCSC RefSeq (source: GRCh37 and GRCh38) - Publicly available gene transcripts from the UCSC Table Browser are used by LDlink’s LDassoc (currently not available in the LDlinkR package), LDmatrix, and LDproxy modules to display genes within the genomic window of interest.

RegulomeDB (source: GRCh37) - Publicly available scores from RegulomeDB are used by LDlink’s LDassoc (currently not available in the LDlinkR package) and LDproxy modules to rank available data types for a single coordinate. GRCh38 support is added via liftOver.

Genetic Map (source: GRCh37) - Publicly available combined recombination rates (cM/Mb) from the 1000 Genomes Project are used by LDlink’s LDassoc (currently not available in the LDlinkR package) and LDproxy modules to show recombination at specific coordinates. GRCh38 support is added via liftOver.

GTEx Portal (source: GRCh38) - Publicly available single-tissue cis-QTL data from the GTEx Portal is used by LDlink’s LDexpress module to show significant variant-gene associations in multiple tissue types. GRCh37 support is added via GTEx lookup table.

GWAS Catalog (source: GRCh38) - Publicly available NHGRI-EBI Catalog of human genome-wide association studies from GWAS Catalog is used by LDlink’s LDtrait module to search if variants have previously been associated with a trait or disease. GRCh37 support is added via dbSNP.

Calculations

LDlink modules report the following measures of linkage disequilibrium: D prime, R squared, and goodness-of-fit statistics. Below is a brief description of each measure.

D prime (D’) - an indicator of allelic segregation for two genetic variants. D’ values range from 0 to 1 with higher values indicating tight linkage of alleles. A D’ value of 0 indicates no linkage of alleles. A D’ value of 1 indicates at least one expected haplotype combination is not observed.

R squared (R²) - a measure of correlation of alleles for two genetic variants. R² values range from 0 to 1 with higher values indicating a higher degree of correlation. An R² value of 0 indicates alleles are independent, whereas an R² value of 1 indicates an allele of one variant perfectly predicts an allele of another variant. R² is sensitive to allele frequency.

Goodness of Fit (X² and p-value) - statistical test testing whether observed haplotype counts follow frequencies expected from variant allele frequencies. High chi-square statistics and low p-values are evidence that haplotype counts deviate from expected values and suggest linkage disequilibrium may be present.

Installation

The release version of LDlinkR can be installed from CRAN:

install.packages("LDlinkR")

The development version of the LDlinkR package can be installed from the GitHub repository by using the remotes package:

install.packages("remotes")
remotes::install_github("CBIIT/LDlinkR")

LDlinkR depends on the following packages:

utils, version 3.4.2 or later
httr, version 1.4.0 or later

Following installation, attach the LDlinkR package with:

library(LDlinkR)

Personal Access Token - Required

In order to access the LDlink API via LDlinkR, we use a personal access token. This is a common convention followed by many APIs and emulates the more familiar HTTPS username/password or SSH keys.

You will need to:

Make a one-time request for your personal access token from a web browser at https://ldlink.nih.gov/?tab=apiaccess.
Once registered, your personal access token will be emailed to you. It is a string of 12 random letters and numbers.
Provide your token as an argument when using LDlinkR. See example below:

LDhap(snps = c("rs3", "rs4", "rs148890987"), 
      pop = "YRI", 
      token = "YourTokenHere123")

Optional:
However, the best security practice is to store your personal access token as an environment variable where LDlinkR can find it and use it on your behalf but where it will not be accidentally shared with the public. Note: Modifying R startup files (such as the .Renviron) is for the advanced R user only. Modification of these files in the wrong way could cause problems. Please proceed cautiously. Step-by-step instructions follow:

After retrieving your personal access token from your email, put your token in your .Renviron file. .Renviron is a hidden file that lives in your home directory. The easiest way to both find and edit the .Renviron file is with a function from the usethis package. From the R console, do:

usethis::edit_r_environ()

Your .Renviron file should open in your editor. Add a line that looks like this:

LDLINK_TOKEN=YourTokenHere123

Important, ensure you put a line break at the end by hitting the enter/return key.

Save and close the .Renviron file. Restart R, as environment variables are only loaded from .Renviron at the start of a new R session. Now, check to see that your token is available by entering:

Sys.getenv("LDLINK_TOKEN")

## [1] "YourTokenHere123"

You should see your personal access token print to the screen, as shown above. Now, LDlinkR function calls that use

Sys.getenv("LDLINK_TOKEN")

for the token argument in LDlinkR function calls will use your personal access token in a private and secure way. This method will be used in the extended examples that follow.

Functions and Examples

LDexpress

Function

LDexpress(snps, 
          pop = "CEU", 
          tissue = "ALL", 
          r2d = "r2", 
          r2d_threshold = 0.1, 
          p_threshold = 0.1, 
          win_size = 500000,
          genome_build = "grch37",
          token = NULL, 
          file = FALSE
         )

Search if a list of genomic variants (or variants in LD with those variants) is associated with gene expression in tissues of interest. Quantitative trait loci data is downloaded from the GTEx Portal.

Arguments

snps, between 1 - 10 variants, using an rsID or chromosome coordinate (e.g. “chr7:24966446”)
pop, a 1000 Genomes Project population, (e.g. YRI or CEU), multiple allowed, default = “CEU”. See the list_pop function in the utilities section below for available human populations and their abbreviation codes.
tissue, select from 1 - 54 non-diseased tissue sites collected for the GTEx project, multiple allowed. Acceptable user input is taken either from “tissue_name_ldexpress” or “tissue_abbrev_ldexpress” (tissue abbreviation) code listed in available GTEx tissue sites using the list_gtex_tissues function (e.g. “ADI_SUB” for Adipose Subcutaneous). Input is case sensitive. Default = “ALL” for all available tissue types.
r2d, Select either “r2” for LD R² (R-squared) or “d” for LD D’, default = “r2”.
r2d_threshold, R-squared or D’ (depends on ‘r2d’ user input parameter) threshold for LD filtering. Any variants within -/+ of the specified genomic window and R² or D’ less than the threshold will be removed. Value needs to be in the range 0 to 1. Default value is 0.1.
p_threshold, define the eQTL significance threshold used for returning query results. Default value is 0.1 which returns all GTEx eQTL associations with P-value less than 0.1.
win_size, set genomic base pair window size for LD calculation. Specify a value greater than or equal to zero and less than or equal to 1000000 base pairs (bp). Default value is -/+ 500000 bp.
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE.

Usage: Single query variant, multiple populations, multiple tissue types using tissue abbreviation

my_output <- LDexpress(snps = "rs4",
                       pop = c("YRI", "CEU"),
                       tissue =  c("ADI_SUB", "ADI_VIS_OME"),
                       win_size = "500000",
                       token = Sys.getenv("LDLINK_TOKEN")
                      )

In the above example, output is a data frame stored in the variable my_output. See below.

head(my_output)

##   Query      RS_ID Position_grch37                R2                D'  Gene_Symbol        Gencode_ID
## 1   rs4 rs10637519  chr13:32430479 0.174249321651574 0.965976331360947 RP1-257C22.2 ENSG00000279314.1
## 2   rs4 rs10637519  chr13:32430479 0.174249321651574 0.965976331360947 RP1-257C22.2 ENSG00000279314.1
## 3   rs4   rs473641  chr13:32431244 0.174249321651574 0.965976331360947 RP1-257C22.2 ENSG00000279314.1
## 4   rs4   rs473641  chr13:32431244 0.174249321651574 0.965976331360947 RP1-257C22.2 ENSG00000279314.1
## 5   rs4   rs671746  chr13:32431263 0.174249321651574 0.965976331360947 RP1-257C22.2 ENSG00000279314.1
## 6   rs4   rs671746  chr13:32431263 0.174249321651574 0.965976331360947 RP1-257C22.2 ENSG00000279314.1
##                         Tissue Non_effect_Allele_Freq Effect_Allele_Freq Effect_Size     P_value
## 1 Adipose - Visceral (Omentum)                G=0.565          GTC=0.435    0.207161  1.0227e-05
## 2       Adipose - Subcutaneous                G=0.565          GTC=0.435    0.225642  2.2578e-07
## 3 Adipose - Visceral (Omentum)                A=0.565            G=0.435    0.207161  1.0227e-05
## 4       Adipose - Subcutaneous                A=0.565            G=0.435    0.225642  2.2578e-07
## 5 Adipose - Visceral (Omentum)                C=0.565            T=0.435    0.207161  1.0227e-05
## 6       Adipose - Subcutaneous                C=0.565            T=0.435    0.226558 1.93289e-07

Usage: Multiple query variants, single population, a tissue type using the full LDexpress tissue name, no spaces, and genome build GRCh38.

my_output <- LDexpress(snps = c("rs345", "rs456"),
                       pop = "YRI",
                       tissue =  "Adipose_Visceral_Omentum",
                       genome_build = "grch38",
                       token = Sys.getenv("LDLINK_TOKEN")
                      )

In the above example, output is a data frame stored in the variable my_output. See below.

head(my_output)

##   Query      RS_ID Position_grch38                R2  D' Gene_Symbol        Gencode_ID
## 1 rs345 rs12877069  chr13:32430415 0.222088835534214  1 RP1-257C22.2 ENSG00000279314.1
## 2 rs345 rs10637519  chr13:32430479  0.10989010989011  1 RP1-257C22.2 ENSG00000279314.1
## 3 rs345   rs473641  chr13:32431244  0.10989010989011  1 RP1-257C22.2 ENSG00000279314.1
## 4 rs345   rs671746  chr13:32431263  0.10989010989011  1 RP1-257C22.2 ENSG00000279314.1
## 5 rs345  rs9315146  chr13:32432193 0.222088835534214  1 RP1-257C22.2 ENSG00000279314.1
## 6 rs345   rs657190  chr13:32432232 0.107871720116618  1 RP1-257C22.2 ENSG00000279314.1
                           Tissue Non_effect_Allele_Freq Effect_Allele_Freq Effect_Size     P_value
## 1 Adipose - Visceral (Omentum)                C=0.685            T=0.315    0.355769 6.11598e-05
## 2 Adipose - Visceral (Omentum)                G=0.519          GTC=0.481    0.207161  1.0227e-05
## 3 Adipose - Visceral (Omentum)                A=0.519            G=0.481    0.207161  1.0227e-05
## 4 Adipose - Visceral (Omentum)                C=0.519            T=0.481    0.207161  1.0227e-05
## 5 Adipose - Visceral (Omentum)                A=0.685            G=0.315    0.276884 2.20517e-08
## 6 Adipose - Visceral (Omentum)                T=0.514            C=0.486    0.207916 9.95318e-06

LDhap

Function

LDhap(snps, 
      pop = "CEU", 
      token = NULL, 
      file = FALSE, 
      table_type = "haplotype",
      genome_build = "grch37"
     )

Calculates population specific haplotype frequencies of all haplotypes observed for a list of query variants. Input is a list of variant RS numbers (concatenated list) and a population group.

Arguments

snps, a list of between 1 - 30 variants, using an rsID or chromosome coordinate (e.g. “chr7:24966446”)
pop, a 1000 Genomes Project population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE
table_type, Choose from one of four options available to determine output format type…haplotype, variant, both and merged. Default = “haplotype”.
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage: Multiple query variants, single population, and genome build GRCh38 High Coverage

LDhap(snps = c("rs3", "rs4", "rs148890987"), 
      pop = "CEU", 
      token = Sys.getenv("LDLINK_TOKEN"),
      genome_build = "grch38_high_coverage"
     )

##   rs3 rs4 rs148890987 Count Frequency
## 1   C   A           C   183    0.9242
## 2   T   G           C    15    0.0758

Usage: Multiple query variants, multiple populations using default genome build GRCh37 (hg19)

LDhap(snps = c("rs3", "rs4", "rs148890987"),
      pop = c("YRI", "CEU"),
      token = Sys.getenv("LDLINK_TOKEN")
     )

##   rs3 rs4 rs148890987 Count Frequency
## 1   C   A           C   355    0.8575
## 2   T   G           C    41     0.099
## 3   T   G           T    11    0.0266
## 4   C   A           T     7    0.0169

Output is a table of alleles, haplotype count and haplotype frequencies.

Usage: Multiple query variants, single population, ‘merged’ output format type.

LDhap(snps = c("rs660670", "rs556780", "rs355", "rs356", "rs542746"),
      pop = "CEU",
      token = Sys.getenv("LDLINK_TOKEN"),
      table_type = "merged",
      genome_build = "grch38"
     )

##   RS_Number Position_grch38    Allele_Frequency Haplotypes       
## 1  rs660670  chr13:31863887    A=0.924, G=0.076          A      G
## 2  rs556780  chr13:31863023    G=0.924, A=0.076          G      A
## 3     rs355  chr13:31883842    A=0.924, G=0.076          A      G
## 4     rs356  chr13:31884663    T=0.924, A=0.076          T      A
## 5  rs542746  chr13:31860055    G=0.924, A=0.076          G      A
## 6                               Haplotype_Count        183     15
## 7                           Haplotype_Frequency     0.9242 0.0758

Output is a table with query variants, genomic position GRCH38 (hg38), etc.

Usage: Multiple query variants, single population, ‘both’ output format type.

LDhap(snps = c("rs660670", "rs556780", "rs355", "rs356", "rs542746"),
      pop = "CEU",
      token = Sys.getenv("LDLINK_TOKEN"),
      table_type = "both"
     )

## [[1]]
##   RS_Number Position_grch37 Allele_Frequency
## 1  rs660670  chr13:32438024 A=0.924, G=0.076
## 2  rs556780  chr13:32437160 G=0.924, A=0.076
## 3     rs355  chr13:32457979 A=0.924, G=0.076
## 4     rs356  chr13:32458800 T=0.924, A=0.076
## 5  rs542746  chr13:32434192 G=0.924, A=0.076
## 
## [[2]]
##   rs660670 rs556780 rs355 rs356 rs542746 Count Frequency
## 1        A        G     A     T        G   183    0.9242
## 2        G        A     G     A        A    15    0.0758

Output is a list that contains both the ‘variant’ and ‘haplotype’ output format types.

LDmatrix

Function

LDmatrix(snps, 
         pop = "CEU", 
         r2d = "r2", 
         token = NULL, 
         file = FALSE,
         genome_build = "grch37"
        )

Generates a data frame of pairwise linkage disequilibrium statistics. Input is a list of between 2 to 2500 variants. Desired output can be based on estimates of R² or D’.

Arguments

snps, list of between 2 - 2500 variants, using an rsID or chromosome coordinate (GRCh37/hg19) (e.g. “chr7:24966446”)
pop, a 1000 Genomes Project population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
r2d, use either “r2” for pairwise R² statistics or “d” for pairwise D’ statistics
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage: Multiple query variants, single population, R² and genome build GRCh38 (hg38).

LDmatrix(snps = c("rs496202", "rs11147477", "rs201578600"), 
         pop = "YRI", 
         r2d = "r2", 
         token = Sys.getenv("LDLINK_TOKEN"),
         genome_build = "grch38"
        )

##     RS_number rs496202 rs11147477 rs201578600
## 1    rs496202    1.000      0.503       0.659
## 2  rs11147477    0.503      1.000       0.786
## 3 rs201578600    0.659      0.786       1.000

Usage: Multiple query variants (mixed use of rsID’s & genomic coordinates), multiple populations, D’

LDmatrix(snps = c("chr13:32444611", "rs11147477", "rs201578600"), 
         pop = c("YRI", "CEU"), 
         r2d = "d", 
         token = Sys.getenv("LDLINK_TOKEN")
        )

##     RS_number rs496202 rs11147477 rs201578600
## 1    rs496202    1.000      0.738       0.973
## 2  rs11147477    0.738      1.000       0.971
## 3 rs201578600    0.973      0.971       1.000

Usage: Multiple query variants read from text file, multiple populations, D’

my_variants <- read.table("variant_list.txt")
my_variants

##           V1
## 1      rs456
## 2      rs114
## 3      rs127
## 4  rs7805287
## 5 rs60676332
## 6 rs10239961

Then, call LDmatrix with:

LDmatrix(snps = my_variants[,1], 
         pop = c("YRI", "CEU"), r2d = "d", 
         token = Sys.getenv("LDLINK_TOKEN")
        )

##   RS_number rs456 rs114 rs127 rs7805287 rs60676332 rs10239961
## 1      rs456 1.000 0.963 0.929     0.789      0.151      1.000
## 2      rs114 0.963 1.000 0.886     0.710      0.148      0.459
## 3      rs127 0.929 0.886 1.000     0.818      0.180      0.912
## 4  rs7805287 0.789 0.710 0.818     1.000      0.094      0.464
## 5 rs60676332 0.151 0.148 0.180     0.094      1.000      0.363
## 6 rs10239961 1.000 0.459 0.912     0.464      0.363      1.000

Output is a table with rows and columns equal to the number of query variants and pairwise linkage disequilibrium statistics.

LDpair

Function

LDpair(var1, 
       var2, 
       pop = "CEU", 
       token = NULL, 
       output = "table", 
       file = FALSE,
       genome_build = "grch37"
      )

Investigates potentially correlated alleles for a pair of variants. Input is two query variants and a 1000 Genomes Project reference population(s) of interest.

Arguments

var1, the first RS number (rsID) or genomic coordinate (GRCh37/hg19) (e.g. “chr7:24966446”), must match a bi-allelic variant
var2, the second RS number or genomic coordinate, as above, must match a bi-allelic variant
pop, a 1000 Genomes Project reference population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
output, two output format options are available, “text”, which displays a two-by-two matrix displaying haplotype counts and allele frequencies along with other statistics, or “table”, which displays the same data in rows and columns, default = “table”
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage: With `output` argument set to “text” and genome build GRCh38 (hg38)

LDpair(var1 = "rs496202", 
       var2 = "rs11147477", 
       pop = "YRI", 
       token = Sys.getenv("LDLINK_TOKEN"), 
       output = "text",
       genome_build = "grch38"
      )

## Query SNPs:
## rs496202 (chr13:32444611)
## rs11147477 (chr13:32509120)
## 
## YRI Haplotypes:
##                rs11147477
##                C       T
##              -----------------
##            C | 11    | 26    | 37    (0.171)
## rs496202     -----------------
##            G | 173   | 6     | 179   (0.829)
##              -----------------
##                184     32      216
##               (0.852) (0.148)
## 
##           G_C: 173 (0.801)
##           C_T: 26 (0.12)
##           C_C: 11 (0.051)
##           G_T: 6 (0.028)
## 
##           D': 0.7737
##           R2: 0.5037
##       Chi-sq: 108.8005
##      p-value: <0.0001
## 
## rs496202(C) allele is correlated with rs11147477(T) allele
## rs496202(G) allele is correlated with rs11147477(C) allele

Usage: With no `output` argument option specified, using default “table”.

LDpair(var1 = "rs496202", 
       var2 = "rs11147477", 
       pop = "YRI", 
       token = Sys.getenv("LDLINK_TOKEN"),
       genome_build = "grch38"
      )

##       var1       var2 pops       var1_pos       var2_pos var1_a1 var1_a2 var1_a1_freq var1_a2_freq var2_a1
## 1 rs496202 rs11147477  YRI chr13:31870474 chr13:31934983       C       G        0.173        0.827       C
##   var2_a2 var2_a1_freq var2_a2_freq d_prime    r2   chisq p_val
## 1       T         0.85         0.15  0.7733 0.503 107.638 1e-04
##                                           corr_alleles
## 1 rs496202(C)-rs11147477(T), rs496202(G)-rs11147477(C)

Output of the output argument “text” option is a two-by-two contingency table displaying haplotype counts and allele frequencies of the two query variants. Also displayed are calculated metrics of linkage disequilibrium including: D prime (D’), R square (R²), and goodness-of-fit (Chi-square and p-value). Goodness-of-fit tests for deviations of expected haplotype frequencies based on allele frequencies. Correlated alleles are reported if linkage disequilibrium is present (R² > 0.1). If linkage equilibrium, no alleles are reported.

Output from the output argument “table” option converts the data from the two-by-two contingency table into a data frame.

LDpop

Function

LDpop(var1, 
      var2, 
      pop = "CEU", 
      r2d = "r2", 
      token = NULL, 
      file = FALSE,
      genome_build = "grch37"
     )

Investigates allele frequencies and linkage disequilibrium patterns across 1000G populations.

Arguments

var1, the first RS number (rsID) or genomic coordinate (GRCh37/hg19) (e.g. “chr7:24966446”), must match a bi-allelic variant
var2, the second RS number or genomic coordinate, as above, must match a bi-allelic variant
pop, a 1000 Genomes Project reference population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
r2d, use “r2” if desired output is based on estimated R² or “d” if D’
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage

LDpop(var1 = "rs496202", 
      var2 = "rs11147477", 
      pop = "YRI", 
      r2d = "r2", 
      token = Sys.getenv("LDLINK_TOKEN"),
      genome_build = "grch38_high_coverage"
     )

##                  Population Abbrev   N rs496202_Allele_Freq rs11147477_Allele_Freq     R2     D'    Chisq P
## 1 Yoruba in Ibadan, Nigeria    YRI 108 G: 82.87%, C: 17.13%   C: 85.19%, T: 14.81% 0.5037 0.7737 108.8005 0

LDproxy

Function

LDproxy(snp, 
        pop = "CEU", 
        r2d = "r2", 
        token = NULL, 
        file = FALSE,
        genome_build = "grch37",
        win_size = "500000"
       )

Explore proxy and putative functional variants for a single query variant. Input is a single RS number and a population group. Depending on the number of query populations, this function could take some time to run.

Arguments

snp, an RS number (rsID) or chromosome coordinate (GRCh37/hg19) (e.g. “chr7:24966446”), one per query, RS number must match a bi-allelic variant
pop, a 1000 Genomes Project reference population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
r2d, use “r2” if desired output is based on estimated R² or “d” if D’
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).
win_size Set base pair (bp) window size. Specify a value greater than or equal to zero and less than or equal to 1,000,000bp. Default value is 500,000bp.

Usage: single reference population and default genome build GRCh37 (hg19).

my_proxies <- LDproxy(snp = "rs456", 
                      pop = "YRI", 
                      r2d = "r2", 
                      token = Sys.getenv("LDLINK_TOKEN")
                     )

Output is a data frame stored in the variable my_proxies with 2455 rows and 10 columns with data.

head(my_proxies)

##    RS_Number         Coord Alleles    MAF Distance Dprime     R2
## 1      rs456 chr7:24962419   (G/C) 0.1944        0      1 1.0000
## 2      rs457 chr7:24962426   (T/C) 0.1944        7      1 1.0000
## 3 rs28475742 chr7:24964633   (G/T) 0.1944     2214      1 1.0000
## 4      rs123 chr7:24966446   (C/A) 0.1944     4027      1 1.0000
## 5      rs125 chr7:24959703   (C/T) 0.2037    -2716      1 0.9436
## 6      rs128 chr7:24958977   (C/T) 0.2037    -3442      1 0.9436
##   Correlated_Alleles RegulomeDB Function
## 1            G=G,C=C          4     <NA>
## 2            G=T,C=C         2b     <NA>
## 3            G=G,C=T          4     <NA>
## 4            G=C,C=A         1f     <NA>
## 5            G=C,C=T         3a     <NA>
## 6            G=C,C=T          7     <NA>

Includes information on all variants within the specified window size of the query variant with a pairwise R² value greater than 0.01.

LDproxy_batch

Function

LDproxy_batch(snp, 
              pop = "CEU", 
              r2d = "r2", 
              token = NULL, 
              append = FALSE,
              genome_build = "grch37",
              win_size = "500000"
             )

Query LDproxy using a list of query variants. LDproxy_batch will make sequential queries, one query per variant. Concurrent queries are not permitted by the LDlink API. Output is saved as text file(s) to the current working directory. Depending on the number of query variants and reference populations selected, this function could take some time to run.

Arguments

snp, a character string or data frame listing RS numbers (rsID) or chromosome coordinates (GRCh37/hg19) (e.g. “chr7:24966446”), one per line.
pop, a 1000 Genomes Project reference population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
r2d, use “r2” if desired output is based on estimated R² or “d” if D’
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
append, a logical, if TRUE, output for each query variant is appended to a single text file and saved to the current working directory. If FALSE, output for each query variant is saved in its own text file with the query variant as the filename. Default value is FALSE.
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).
win_size Set base pair (bp) window size. Specify a value greater than or equal to zero and less than or equal to 1,000,000bp. Default value is 500,000bp.

Usage: multiple variants, default `pop` and `r2d`

The list of query variants passed to LDproxy_batch can be stored as a character string.

LDproxy_batch(snp = c("rs456", "rs114", "rs127"), 
              token = Sys.getenv("LDLINK_TOKEN")
             )

Or, a longer list of variants can be read into a data frame from a text file and passed into LDproxy_batch. The list should be in a simple text file, one query variant per line. For example:

my_variants <- read.table("variant_list.txt")
my_variants

##           V1
## 1      rs456
## 2      rs114
## 3      rs127
## 4  rs7805287
## 5 rs60676332
## 6 rs10239961

Then, call LDproxy_batch with:

LDproxy_batch(snp = my_variants, 
              token = Sys.getenv("LDLINK_TOKEN")
             )

Output not displayed. All output from LDproxy_batch is saved to a text file(s) in the current working directory.

LDtrait

Function

LDtrait(snps,
        pop = "CEU",
        r2d = "r2",
        r2d_threshold = 0.1,
        win_size = 500000,
        token = NULL,
        file = FALSE,
        genome_build = "grch37"
       )

Search if a list of variants (or variants in LD with those variants) have been previously associated with a trait or disease. Trait and disease data is updated nightly from the GWAS Catalog.

Arguments

snps, between 1 - 50 variants, using an rsID or chromosome coordinate (GRCh37)(e.g. “chr7:24966446”). All input variants must match a bi-allelic variant.
pop, a 1000 Genomes Project population, (e.g. YRI or CEU), multiple allowed, default = “CEU”. See the list_pop function in the utilities section below for available human populations and their abbreviation codes.
r2d, use “r2” to filter desired output from a threshold based on estimated LD R2 (R squared) or “d” for LD D’ (D-prime), default = “r2”.
r2d_threshold, R-squared or D’ (depends on ‘r2d’ user input parameter) threshold for LD filtering. Any variants within -/+ of the specified genomic window and R² or D’ less than the threshold will be removed. Value needs to be in the range 0 to 1. Default value is 0.1.
win_size, set genomic base pair window size for LD calculation. Specify a value greater than or equal to zero and less than or equal to 1000000 base pairs (bp). Default value is -/+ 500000 bp.
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site..
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE.
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage: Single query variant, multiple reference populations and genome build GRCh38(hg38)

LDtrait(snps = "rs456",
        pop = c("YRI", "CEU"),
        token = Sys.getenv("LDLINK_TOKEN"),
        genome_build = "grch38"
       )

The following is the output from the above function call.

##   Query                      GWAS_Trait  RS_Number Position_GRCh38          Alleles
## 1 rs456 Highest math class taken (MTAG) rs10248878   chr7:24869118 C=0.175, T=0.825
## 2 rs456   Educational attainment (MTAG)      rs457   chr7:24922807 C=0.697, T=0.303
##                 R2                D' Risk_Allele Effect_Size_95_CI    Beta_or_OR P_value
## 1 0.41175133337888 0.920247773906311      0.5967            0.0104 0.0071-0.0137   7e-10
## 2                1                 1      0.4495            0.0072 0.0047-0.0097   4e-08

Usage: Multiple query variants, multiple reference populations and win_size set to 750000 base pairs (bp), default genome build GRCh37(hg19).

LDtrait(snps = c("rs114", "rs496202", "rs345"),
        pop = c("YRI", "CHB", "CEU"),
        win_size = "750000",
        token = Sys.getenv("LDLINK_TOKEN")
       )

Output of the above function is below.

##      Query                                        GWAS_Trait  RS_Number
## 1    rs114                   Highest math class taken (MTAG) rs10248878
## 2    rs114                     Educational attainment (MTAG)      rs457
## 3 rs496202                                  Refractive error      rs353
## 4    rs345            DNA methylation variation (age effect)   rs203425
## 5    rs345 Facial morphology (factor 14, intercanthal width)   rs799522
##   Position_GRCh37          Alleles                R2                D'
## 1   chr7:24908737 C=0.123, T=0.877 0.200231693692643 0.897255733792921
## 2   chr7:24962426 C=0.748, T=0.252  0.56312684849231 0.969967060647161
## 3  chr13:32454349 A=0.902, G=0.098                 1                 1
## 4  chr13:32468087 A=0.074, T=0.926 0.954994192799071                 1
## 5  chr13:32514028 C=0.769, T=0.231 0.236284178064096 0.918763102725367
##   Risk_Allele Effect_Size_95_CI    Beta_or_OR P_value
## 1      0.5967            0.0104 0.0071-0.0137   7e-10
## 2      0.4495            0.0072 0.0047-0.0097   4e-08
## 3        <NA>              <NA>          <NA>   1e-12
## 4          NR              <NA>          <NA>   2e-08
## 5      0.1263            0.2157     0.12-0.31   6e-06

SNPchip

Function

SNPchip(snps, 
        chip = "ALL", 
        token = NULL, 
        file = FALSE,
        genome_build = "grch37"
       )

Used to find commercial genotyping chip arrays for variants. Input is a list of between 1 - 5000 variants (one per line) and desired commercial chip arrays to search. Input variants do not need to be on the same chromosome.

Arguments

snps, between 1 - 5,000 variants, using an rsID or chromosome coordinate (e.g. “chr7:24966446”)
chip, chip or arrays, platform code(s) for a SNP chip array, ALL_Illumina, ALL_Affy or ALL, default=ALL, use the list_chips utility (see below) to lookup available commercial SNP chip arrays and their codes.
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE.
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage: Multiple variants, search “ALL” available chip arrays

SNPchip(snps = c("rs3", "rs4", "rs148890987"), 
        chip = "ALL", 
        token = Sys.getenv("LDLINK_TOKEN")
       )

## WARNING: The following RS number did not have any platforms found: rs148890987, rs3.

##     RS_Number Position_GRCh37 A_SNP5.0 A_CHB2 A_250S A_SNP6.0
## 1 rs148890987  chr13:32403784        0      0      0        0
## 2         rs3  chr13:32446842        0      0      0        0
## 3         rs4  chr13:32447222        1      1      1        1

Usage: Multiple variants, search two Affymetrix arrays, and genome build GRCh38 (hg38)

SNPchip(snps = c("rs3", "rs4", "rs148890987"), 
        chip = c("A_SNP5.0", "A_CHB2"), 
        token = Sys.getenv("LDLINK_TOKEN"),
        genome_build = "grch38"
       )

## WARNING: The following RS number did not have any platforms found: rs148890987, rs3.

##     RS_Number Position_GRCh38 A_SNP5.0 A_CHB2
## 1 rs148890987     13:31829647        0      0
## 2         rs3     13:31872705        0      0
## 3         rs4     13:31873085        1      1

Usage: Multiple variants, search all available Affymetrix arrays using, “ALL_Affy”

SNPchip(snps = c("rs3", "rs4", "rs148890987"), 
        chip = "ALL_Affy", 
        token = Sys.getenv("LDLINK_TOKEN")
       )

## WARNING: The following RS number did not have any platforms found: rs148890987, rs3.

##     RS_Number Position_GRCh37 A_SNP5.0 A_CHB2 A_250S A_SNP6.0
## 1 rs148890987  chr13:32403784        0      0      0        0
## 2         rs3  chr13:32446842        0      0      0        0
## 3         rs4  chr13:32447222        1      1      1        1

Output is a data frame of query variant rows (RS number), genomic coordinate (GRCh37) and genotyping chip array columns. The presence of a “1” designates the variant is present on the respective commercial genotyping array and a “0” indicates that it is not present on the genotyping array.

SNPclip

Function

SNPclip(snps, 
        pop = "CEU",
        r2_threshold = "0.1", 
        maf_threshold = "0.01", 
        token = NULL, 
        file = FALSE,
        genome_build = "grch37"
       )

Prune a list of variants by linkage disequilibrium. Input is a list of variant RS numbers (one per line) and a population group.

Arguments

snps, a list of between 1 - 5,000 variants, using an RS number (rsID) or chromosome coordinate (GRCh37) (e.g. “chr7:24966446”). All input variants must be on the same chromosome and match a bi-allelic variant.
pop, a 1000 Genomes Project reference population, uses three letter population code, (e.g. YRI or CEU), multiple allowed, default = “CEU”
r2_threshold, Used to set the R² threshold for LD pruning. One of each pair of variants with a R² greater than the threshold is removed. Value needs to be in the range 0 to 1. Default value is 0.1.
maf_threshold, Used to set minor allele frequency (MAF) threshold for LD pruning. Variants with a MAF less than or equal to the threshold are removed. Value needs to be in the range 0 to 1. Default value is 0.01.
token, LDlink provided user access token is required, default = NULL, register for a free token on the LDlink web site.
file, optional character string naming a path and file for saving results. If file = FALSE, no file will be generated, default = FALSE.
genome_build Choose between one of three options…grch37 for genome build GRCh37 (hg19), grch38 for GRCh38 (hg38), or grch38_high_coverage for GRCh38 High Coverage (hg38) 1000 Genome Project data sets. Default is GRCh37 (hg19).

Usage: Multiple Variants

SNPclip(snps =  c("rs3", "rs4", "rs148890987", "rs115955931"), 
        pop = "YRI", 
        r2_threshold =  "0.1", 
        maf_threshold = "0.01", 
        token = Sys.getenv("LDLINK_TOKEN"),
        genome_build = "grch37"
       )

##     RS_Number       Position          Alleles
## 1         rs3 chr13:32446842 C=0.829, T=0.171
## 2         rs4 chr13:32447222 A=0.829, G=0.171
## 3 rs148890987 chr13:32403784     C=1.0, T=0.0
## 4 rs115955931 chr13:32130008 G=0.954, A=0.046
##                                             Details
## 1                                     Variant kept.
## 2 Variant in LD with rs3 (R2=1.0), variant removed.
## 3              Variant MAF is 0.0, variant removed.
## 4                                     Variant kept.

The output table provides details including query variant RS number, genomic position, alleles, and and details about whether the variant was kept or removed.

Utilities and Examples

list_chips

Function

list_chips()

Provides a data frame listing the names and abbreviation codes for available commercial SNP Chip Arrays from Illumina and Affymetrix.

Usage

list_chips()

list_pop

Function

list_pop()

Provides a data frame listing the available reference populations from the 1000 Genomes Project, continental or super-populations (e.g. European, African, Admixed American) and sub-populations (e.g Finnish, Gambian, Peruvian)

Usage

list_pop()

list_gtex_tissues

Function

list_gtex_tissues()

Provides a data frame listing the GTEx full names, LDexpress full names (without spaces) and acceptable abbreviation codes of the 54 non-diseased tissue sites collected for the GTEx Portal and used as input for the LDexpress function.

Usage

list_gtex_tissues()

FAQs (Frequently Asked Questions)

What if my access token doesn’t work?
- Please double check that the token was typed accurately. Then, ensure the format of the function call is correct. For example, if your alphanumeric access token is: 123abc456789, then, use it as:
```
df <- LDproxy(snp = "rs456", pop = "YRI", token = "123abc456789")
```

If you still can not solve the problem, please email us at NCILDlinkWebAdmin@mail.nih.gov.

Can I set a threshold or cut-off value for R² or D` values?
- No. LDlinkR functions do not include ‘threshold’ as an argument. However, the returned data object can be subset using base R. For example:

df <- LDproxy("rs12027135", pop = "CEU",r2d = "r2", token = "YourTokenHere123")
new_df <- subset(df, R2 >= 0.8)

I need to upload hundreds of variants from a text file into LDmatrix. Why do I get an error with the following code?

test <- read.table("variant_list.txt", header = FALSE)
LDmatrix(snps = test, pop = "CEU", r2d = "r2", token = "YourTokenHere123")

Error in LDmatrix(snps = test, pop = "CEU", r2d = "r2", token = "YourTokenHere123"), : Input is between 2 to 1000 variants.

In the above example, ‘test’ is an object of class data frame and the variants are in the first column. The first column needs to be specified in order to avoid the error. The example below will yield the desired result:

test <- read.table("variant_list.txt", header = FALSE)
LDmatrix(snps = test[,1], pop = "CEU", r2d = "r2", token = "YourTokenHere123")

##    RS_number rs60676332 rs7805287 rs127 rs456 rs10239961 rs114
## 1 rs60676332      1.000     0.008 0.013 0.017      0.286 0.039
## 2  rs7805287      0.008     1.000 0.980 0.882      0.170 0.614
## 3      rs127      0.013     0.980 1.000 0.900      0.167 0.632
## 4      rs456      0.017     0.882 0.900 1.000      0.177 0.722
## 5 rs10239961      0.286     0.170 0.167 0.177      1.000 0.008
## 6      rs114      0.039     0.614 0.632 0.722      0.008 1.000

What genome build does LDlink use for genomic coordinates?
- All genomic coordinates are based on GRCh37/hg19.

How can I ask for help?
- If you find that you can’t answer a question or solve a problem yourself, you can email us at NCILDlinkWebAdmin@mail.nih.gov or open an issue at the LDlinkR GitHub repository (https://github.com/CBIIT/LDlinkR).

Session Information

sessionInfo()

## R version 4.3.2 (2023-10-31)
## Platform: x86_64-apple-darwin20 (64-bit)
## Running under: macOS Sonoma 14.4.1
## 
## Matrix products: default
## BLAS:   /Library/Frameworks/R.framework/Versions/4.3-x86_64/Resources/lib/libRblas.0.dylib 
## LAPACK: /Library/Frameworks/R.framework/Versions/4.3-x86_64/Resources/lib/libRlapack.dylib;  LAPACK version 3.11.0
## 
## locale:
## [1] C/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
## 
## time zone: America/New_York
## tzcode source: internal
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## loaded via a namespace (and not attached):
##  [1] digest_0.6.35     R6_2.5.1          fastmap_1.1.1     xfun_0.43         cachem_1.0.8     
##  [6] knitr_1.45        htmltools_0.5.8.1 rmarkdown_2.26    lifecycle_1.0.4   cli_3.6.2        
## [11] sass_0.4.9        jquerylib_0.1.4   compiler_4.3.2    rstudioapi_0.16.0 tools_4.3.2      
## [16] evaluate_0.23     bslib_0.7.0       yaml_2.3.8        rlang_1.1.3       jsonlite_1.8.8

LDlinkR: An R Package for Rapidly Calculating Linkage Disequilibrium Statistics in Diverse Populations

Timothy A. Myers, Stephen J. Chanock and Mitchell J. Machiela

2024-04-10

Introduction

Understanding Linkage Disequilibrium

LDlink Data Sources

Calculations

Installation

Personal Access Token - Required

Functions and Examples

LDexpress

Function

Arguments

Usage: Single query variant, multiple populations, multiple tissue types using tissue abbreviation

Usage: Multiple query variants, single population, a tissue type using the full LDexpress tissue name, no spaces, and genome build GRCh38.

LDhap

Function

Arguments

Usage: Multiple query variants, single population, and genome build GRCh38 High Coverage

Usage: Multiple query variants, multiple populations using default genome build GRCh37 (hg19)

Usage: Multiple query variants, single population, ‘merged’ output format type.

Usage: Multiple query variants, single population, ‘both’ output format type.

LDmatrix

Function

Arguments

Usage: Multiple query variants, single population, R2 and genome build GRCh38 (hg38).

Usage: Multiple query variants (mixed use of rsID’s & genomic coordinates), multiple populations, D’

Usage: Multiple query variants read from text file, multiple populations, D’

LDpair

Function

Arguments

Usage: With output argument set to “text” and genome build GRCh38 (hg38)

Usage: With no output argument option specified, using default “table”.

LDpop

Function

Arguments

Usage

LDproxy

Function

Arguments

Usage: single reference population and default genome build GRCh37 (hg19).

LDproxy_batch

Function

Arguments

Usage: multiple variants, default pop and r2d

LDtrait

Function

Arguments

Usage: Single query variant, multiple reference populations and genome build GRCh38(hg38)

Usage: Multiple query variants, multiple reference populations and win_size set to 750000 base pairs (bp), default genome build GRCh37(hg19).

SNPchip

Function

Arguments

Usage: Multiple variants, search “ALL” available chip arrays

Usage: Multiple variants, search two Affymetrix arrays, and genome build GRCh38 (hg38)

Usage: Multiple variants, search all available Affymetrix arrays using, “ALL_Affy”

SNPclip

Function

Arguments

Usage: Multiple Variants

Utilities and Examples

list_chips

Function

Usage

list_pop

Function

Usage

list_gtex_tissues

Function

Usage

FAQs (Frequently Asked Questions)

Session Information

Usage: Multiple query variants, single population, R² and genome build GRCh38 (hg38).

Usage: With `output` argument set to “text” and genome build GRCh38 (hg38)

Usage: With no `output` argument option specified, using default “table”.

Usage: multiple variants, default `pop` and `r2d`