cojopy

Conditional Analysis with LD Matrix

Get the same results as GCTA COJO, but with LD matrix.

• Documentation

Installation

pip install cojopy

or with uv:

uv pip install cojopy

Usage

Input files

Summary statistics file

The summary statistics file should have the following columns (same as the cojo file of GCTA):

• SNP: SNP ID
• A1: Effect allele
• A2: Other allele
• b: Effect size
• se: Standard error
• p: P-value
• freq: Minor allele frequency
• N: Sample size

LD matrix file

• A tab-delimited LD matrix, same as the output of plink --r square.

  [!Note] The allele order of the summary statistics file should be the same as the allele order of the LD matrix file.

Stepwise model selection of independent associated SNPs (same as gcta --cojo-slct)

cojo slct \
--sumstats ./exampledata/sim_gwas.ma \
--ld-matrix ./exampledata/sim_r.ld \
--output ./exampledata/slct.txt

Fit all the included SNPs to estimate their joint effects without model selection (same as gcta --cojo-joint)

cojo joint \
--sumstats ./exampledata/sim_gwas.ma \
--ld-matrix ./exampledata/sim_r.ld \
--extract-snps ./exampledata/extract_snps.txt \
--output ./exampledata/joint.txt

Perform association analysis of the included SNPs conditional on the given list of SNPs (same as gcta --cojo-cond)

cojo cond \
--sumstats ./exampledata/sim_gwas.ma \
--ld-matrix ./exampledata/sim_r.ld \
--cond-snps ./exampledata/cond_snps.txt \
--output ./exampledata/cond.txt

Parameters

• cond-snps: For cond command, path to the file containing the SNPs to condition on, one SNP per line.
• extract-snps: For joint and cond commands, path to the file containing the SNPs to extract, one SNP per line.
• ld-freq: Path to the LD frequency file, optional, use freq in sumstats if not provided.
• p-cutoff: P-value cutoff, default is 5e-8.
• collinear-cutoff: Collinearity cutoff, default is 0.9.
• maf-cutoff: Minor allele frequency cutoff, default is 0.01.
• diff-freq-cutoff: Difference in minor allele frequency cutoff, default is 0.2, won't take effect unless ld-freq is provided.