Conditional analysis results

This page has last been updated for R14.

Conditional analysis algorithm

All regions with genome-wide significant results were subjected to Regenie conditional analysis (i.e. exactly the same regions that are also finemapped). The most significant variant is added as a covariate and all the other variants in the region are tested for association, conditionally on the top variant. This process is continued, adding new variants to the list of covariates, until there are no variants with conditional p-value < 1e-6. The same covariates were used as in the main FinnGen core GWAS analysis.

Independent snps file

All independent top SNPs in the region tested. CHR_POS_REF_ALT is the variant id for the most significant SNP in the region in unconditional analysis.

independent.snps file structure

Column	Description
VARIANT	Top SNP of the iteration
BETA	unconditional beta of top SNP
SE	unconditional standard error of top SNP
MLOG10P	unconditional -log10 p-value of top SNP
BETA_cond	conditional beta of top SNP
SE_cond	conditional standard error of top SNP
MLOG10P_cond	conditional -log10 p-value of top SNP
VARIANT_cond	list of variants that were used as conditioning variants

Conditional file

Conditional results for all variants in each iteration in the above summary file. #ITER corresponds to each iteration in the independent snps file. CHR_POS_REF_ALT is the variant id for the most significant SNP in the region in unconditional analysis.

conditional file structure

Column	Description
CHROM	chromosome of tested variant
GENPOS	position of tested variant
ID	CHROM_POS_REF_ALT of tested variant
ALLELE0	Reference allele
ALLELE1	Effect allele
A1FREQ	Effect allele frequency
INFO	Imputation INFO score (IMPUTE method formula as output by REGENIE)
N	sample size
TEST	test performed
BETA	condititional beta
SE	conditional standard error
CHISQ	chisq statistic of association
LOG10P	-log10 p-value
EXTRA	Additional notes by Regenie