WISARD official site

Select O/S : [?]

Case tutorial

Linkage Disequilibrium

This section describes about

Introduction for LD
- Computing r² and D'
- Computing correlation r²
Various types LD limits
- Exhaustive
- Exhaustive chromosome-wise
- Exhaustive window-wise
- Exhaustive gene-wise
- Bin-size
- Fixing some variants in each pair for LD
Genotype clumping with LD

Introduction for LD [top]

Linkage disequilibrium (LD) means the amount of nonrandomness of genotype distribution between two loci. WISARD can calculate r² and D'(in default) ro correlation-based r² (with --ldcor) as LD measures, and several ways to pair variants for LD are available. For instance, we consider the following example data:

Example 1 : Assume dataset formed as this...

1 SNP1  0 129890 (Genotype...)
1 SNP2  0 129950 ...
1 SNP3  0 129960 ...
1 SNP4  0 130020 ...
1 SNP5  0 131000 ...
1 SNP6  0 142729 ...
2 SNP7  0 91284  ...
2 SNP8  0 91300  ...
2 SNP9  0 91800  ...
2 SNP10 0 95520  ...
2 SNP11 0 98582  ...

Computing r² and D'

In default, WISARD computes r² and D' in LD computation, with --ld option.

Computing exhaustive LD, with $r^2$ and $D'$ measure C:\Users\WISARD> wisard --bed test_miss0.bed --ld --out ld_all

Computing correlation r²

In default, WISARD computes r² and D' in LD computation. This can be replaced with correlation r² with --ldcor option, and it can be applied to the options for LD, which are described below.

Computing exhaustive LD, with correlation $r^2$ measure C:\Users\WISARD> wisard --bed test_miss0.bed --ld --ldcor --out ld_all_cor

Various types LD limits [top]

Exhaustive

Exhaustive two-way pairs of variants indicates all possible pairs of variants and if there are m variants, 0.5m(m-1) LDs are calculated. For example data, there are 55 exhaustive two-way pairs. WISARD can generate exhaustive two-way LD by using --ld and this is calculated by default. Note that the computational burden for exhaustive two-way LD quadratically incrase.

Computing exhaustive LD C:\Users\WISARD> wisard --bed test_miss0.bed --ld --out ld_all

Exhaustive chromosome-wise

WISARD can calculate LDs for all pairs of variants in each chromosome. For example data, there are 6 variants in chromosome 1 and 5 variants in chromosome 2, and thus 15+10 LDs are calculated. It can be calculated by adding --ldchr option.

Computing chromosome-wise exhaustive LD C:\Users\WISARD> wisard --bed test_miss0.bed --ld --ldchr --out ld_all

Exhaustive window-wise

Exhaustive window-wise pairs means pairs of variants of which distance(basepair) is less than the specified value. It can be calculated by adding --ldsize options and window-size in basepair should be specified. For example data, if we add --ldsize 1000 option, the following pairs of variants will be calculated:
- SNP1 and SNP1, SNP2, SNP3, SNP4
- SNP2 and SNP2, SNP3, SNP4
- SNP3 and SNP3, SNP4
- SNP4 and SNP4, SNP5
- SNP5 and SNP5
- SNP6 and SNP6
- SNP7 and SNP7, SNP8, SNP9
- SNP8 and SNP8, SNP9
- SNP9 and SNP9
- SNP10 and SNP10
- SNP11 and SNP11

Computing 1K window-wise exhaustive LD C:\Users\WISARD> wisard --bed test_miss0.bed --ld --ldsize 1000 --out ld_all

Bin-size

WISARD can calculate LDs for certain number of consecutive pairs of variants by using --ldbin option. If we add "--ldbin k", LDs between each variant and its k consecutive variants are calculated.

Gene-wise and bin-exhaustive LD computation C:\Users\WISARD> wisard --bed test_miss0.bed --ld --set test_gene_ld.txt --ldbin 2

Under the above command, the following pairs of variants are considered:
- SNP1 and SNP1, SNP2, SNP3
- SNP2 and SNP2, SNP3, SNP4
- SNP3 and SNP3, SNP4, SNP5
- SNP4 and SNP4, SNP5, SNP6
- SNP5 and SNP5, SNP6
- SNP6 and SNP6
- SNP7 and SNP7, SNP8, SNP9
- SNP8 and SNP8, SNP9, SNP10
- SNP9 and SNP9, SNP10, SNP11
- SNP10 and SNP10, SNP11
- SNP11 and SNP11

Fixing some variants in each pair for LD

LDs for pairs including specified variant can be calculated with WISARD by using --ldvar option. Name of variants must be specified as arguments for --ldvar option. For instance, consider the following code for example data:

Calculate LD between SNP1/SNP10 and their adjacent two variants C:\Users\WISARD> wisard --bed test_miss0.bed --ld --ldbin 2 --ldvar SNP1,SNP10

Example codes calculate LDs for the following pairs of variants:
- SNP1 and SNP1, SNP2, SNP3
- SNP10 and SNP10, SNP11

NOTE!

An argument of --ldvar can be a file path which contains variant names per line

Genotype clumping with LD [top]

!!! Experimental function !!!

Highly correlated variants can be simply clumped into single variable with WISARD, and this is also useful for rare variants. WISARD supports this functionality with --makeclgeno option. To define the variants which are clumped into a single variable, additional option is required. Note that an additonal option is required to which threshold is most appropriate to clumping variants into single variable.

Edit this page

Last modified : 2017-09-13 16:30:08