Kanta lab association results v1

The first set of Kanta lab value association results, including GWAS, finemap, autoreporting, and ldsc (heritability and genetic correlations) results for all unique lab measurements (OMOPID) with at least 1,000 individuals with at least one measurement (n = 383 lab measurements) can be found from green library: /finngen/library-green/lab_values/gwas_release_2024_12_1/

These results are in pheweb browser: https://kanta.finngen.fi

QC and data preparation

(see data_documentation/Kanta_labs_over1000_individuals_summary_table.txt for further details on each lab measurement)

• Clear outliers were removed, if any (see Kanta_labs_over1000_individuals_summary_table.txt for detailed outlier filters)

• The distributions of the measurements were looked individually, and analyzed as raw values as possible, with the following numbers:

  • 70 were analyzed as raw median values,

  • 120 were logarithm transformed,

  • 157 were normalized (inverse-rank), and

  • 36 were analyzed as binary (ever having measurement above a certain threshold or test outcome as ‘abnormal’ or ‘high’ vs never having measurement above a certain threshold or test outcome as ‘abnormal’ or ‘high’)

• Sample-wise median was calculated for all lab measurements that were analyzed as quantitative

GWAS

• GWAS was performed for all 383 lab measurements using REGENIE pipeline

• Same covariates (age at end of follow-up or death, sex, genotyping batch and PC1-10) were used as in core endpoint GWASes

• Linear mixed model was used for lab measurements that were analyzed as quantitative, and logistic mixed model was used for binary phenotypes

• Summary statistics were filtered for variants with MAC >100 (for quantitative only) and imputation info score > 0.6

FINEMAP

• All 383 summary statistics were finemapped using finemap pipeline

  • GWS (p < 5E-8) regions were finemapped

  • MHC region was excluded

Autoreporting

• For all finemapped (GWS regions, MHC region excluded) regions, autoreporting was performed using pipeline described here: https://github.com/FINNGEN/autoreporting

LDSC (heritability estimates and genetic correlations)

• For all 383 lab measurements

  • Heritability estimates and

  • Genetic correlations with all R12 endpoints (2,469) were calculated using LDSC pipeline