LD calculation#

GWAS low-rank LD matrix#

Options#

--block-info param-ref4cM_v37.pos#
Read the minimum width (cM) of a LD block from param-ref4cM_v37.pos.
--make-block-ldm#
Generate full LD matrix for each LD blocks.
--ldm ldblockFolder#
Input folder that contains a serise of binary files for blocks block*.ldm.bin, a text file for genome-wide SNP information snp.info, and a text file for block information ldm.info.
--merge-block-ldm-info#
merge LD matrix info files across blocks
--make-ldm-eigen#
perform eigen-decomposition for each block one by one and store the result in binary format block*.eigen.bin in folder ldblockFolder
--make-eigen#
Generate LD block from genotype file

Examples#

Way 1. Generate low-rank LD matrix from blockwise full LD matrix

##Step 1. generate full LD matrices for each LD block
BayesOmics64 --bfile test --make-block-ldm
    --block-info ref4cMFile --out ldblockFolder
## Step 2. after all chromosomes/blocks are done,
## run the command below to merge info files across blocks
BayesOmics64 --ldm ldblockFolder --merge-block-ldm-info \
    --out ldblockFolder
## Step 3. Do eigen-decomposition for each block one by one,
## and store the result in binary format block***.eigen.bin in folder ldm
BayesOmics64 --ldm ldblockFolder --make-ldm-eigen \
     --out ldblockFolder

If you want to learn more about way 1 process, go to SBayesRC tutorial website for details.

Way 2. Generate low-rank LD matrix from genotype directly

BayesOmics64 --bfile test --make-eigen \
    --block-info ref4cMFile \
    --out test

This option will generate three files: a text file that contains ld block information (test.eigen.ldblock.info), a text file that contains SNP information (test.eigen.ldblock.snp.info) and a binary file that contains the full LD matrix data (test.eigen.ldblock.bin).

xQTL low-rank LD matrix#

Options#

LD martrix for molecular region

--make-eigen-gene test_besd#
Compute a low rank LD matrix reference for gene expression from the genotype data provided by –bfile option. This option will generate three files: a text file that contains ld block information (test.eigen.gene.info), a text file that contains SNP information (test.eigen.gene.snp.info) and a binary file that contains the full LD matrix data (test.eigen.gene.bin). The input file is SMR BESD format.
--add-gene-n geneSampleSize.txt#
Manage gene sample size

The content of **geneSampleSize.txt:

ENSG00000183628       10000
ENSG00000100034       10000
......

Columns are gene ensgid id and sample size.

If you have individual sample size for each gene, you can use –add-gene-n flag to add it. if per gene sample size is missing, you can also add total sample size to besd format file by using SMR software, and BayesOmics will use total sample size for each gene. If both total and per gene sample size are missing, BayesOmics will use estimated gene sample size in the analysis.

Examples#

make a low rank LD matrix for gene

BayesOmics64 --bfile test --beqtl-summary mybesd \
    --var-prop 0.9995 --make-eigen-gene  test_besd \
    --out test

To accelerate this process, we provide an flag --keep-one-gene to do LD eigen-decompostion for each gene in parallel and then merge them into corresponding chromosome file.

## genearte low-rank gene LD for given gene
BayesOmics64 --bfile test \
    --beqtl-summary mybesd \
    --var-prop 0.9995 \
    --keep-one-gene geneID \
    --make-eigen-gene --out test

## merge multiple low rank gene LDs based on configure file
BayesOmics64  --eigen-list multi_eigen.txt \
    --merge-eigen-gene \
     --out test
--merge-eigen-gene  multi_eigen.txt#
Input multiple low-rank LD matrix for gene in eigen format (See the option –calc-eigen-gene). The root filenames of multiple low-rank LD matrix are given in a file, e.g. multi_eigen.txt

multi_grm.txt (full paths can be specified if the LD matrix files are in different directories)

test_chr1
test_chr2
test_chr3
......
test_chr22
On this page