Big-LD: A new haplotype block detection method

Sunah Kim (sunny03@snu.ac.kr)

Big-LD

Big-LD is a block partition method based on interval graph modeling of LD bins which are clusters of strong pairwise LD SNPs, not necessarily physically consecutive. The detailed information about the Big-LD can be found in our paper published in bioinformatics.

The Big-LD algorithm and visualization function has been updated and implemented in a new package "gpart". The package "gpart" can be downloaded from bioconductor

The "gpart" has following features.

the updated BigLD function

• supports two types of heuristic algorithm, near-nonhrst and fast.
• supports LD measures r2 and D' both.
• supports PLINK formats (.ped, .map, .raw, .traw) and vcf format as input

the updated LDblockHeatmap function

• shows up to 20000 SNPs.
• shows physical locations of LD blocks and SNPs
• shows gene locations if the gene information is inputted (or by loading gene information from the Ensembl or UCSC database) .

the new SNP sequence partition algorithm GPART

• divides the entire SNP sequence using LD block information (BigLD result) and gene information
• can limit the min/max size of a block

A BigLD result plotted by the updated LDblockHeatmap() in "gpart" package..

• EAS population of 1000 Genomes phase 1 data
• chr1:16Mb-17.9Mb