Skip to content

PPP CT-LDSC (Pulit et al, BMI)

I applied Cross Trait Linkage Disequilibrium Score Regression (CT-LDSC)1 to estimate genetic correlation between the Pullit et al.2 meta-GWAS of body mass index (BMI) and proteomic GWAS from the European discovery cohort of the UK Biobank Pharma Proteomics Project (UKBB PPP)3.

Note that because of its assumptions of uniform polygenicity, CT-LDSC measures genetic correlation due to diffuse polygenic effects. It does not measure genetic correlation due to highly-concentrated locus-specific effects.

Results

As is standard for LDSC analysis, I restricted to the statistics to Hapmap3 variants, and excluded the MHC region. I used the standard thousand genomes linkage disequilibrium scores provided by the authors of LDSC. Because my previous heritability experiment suggested little difference between using all SNPs and excluding the cis-region near the protein of interest, I only ran this experiment with the cis region excluded.

The results are below:

Loading table…

Columns: oid: Olink assay ID; gene: name of gene/protein under study; rg: CT-LDSC genetic correlation estimate; rg_se: jackknife standard error of CT-LDSC genetic correlation estimate; rg_p: p value of test that rg is not zero; gcov: estimated genetic covariance; inter: intercept term in CT-LDSC regression; h2_trait: trait heritability estimate; h2_prot: protein heritability estimate; n_snps: number of hapmap3 variants included; spr: for cases in which multiple rows corresponding to distinct Olink assays of the same protein have been merged into a single row, this gives the maximum spread between the rg values of the merged rows; s_bh: True if the null hypothesis is rejected under the Benjamini-Hochberg procedure at an FDR of 0.05; s_bon: True if the null hypothesis is rejected under the Bonferroni correction at a significance level of 0.05.

Interpretation

Many pharma proteomics project proteins are significantly genetically correlated with BMI. Moreover, the most strongly correlated proteins appear to be consistent with the existing literature

For instance, the most strongly genetically correlated protein is leptin (LEP), which is known to produced by fat cells in proportion to fat mass4. It is also interesting to observe the strong correlation with complement-pathway-related proteins like complement factor B (CFB).

Recall that a significant genetic correlation tells us that there is a shared genetic architecture between BMI and the plasma level of a protein of interest. However, genetic correlation does not inform us about the causal relationship between the protein and BMI. To draw causal conclusions, we would need a causal inference technique like Mendelian Randomization5 or LCV6.

  1. Brendan Bulik-Sullivan, Hilary K Finucane, Verneri Anttila, Alexander Gusev, Felix R Day, Po-Ru Loh, ReproGen Consortium, Psychiatric Genomics Consortium, Genetic Consortium for Anorexia Nervosa of the Wellcome Trust Case Control Consortium 3, Laramie Duncan, and others. An atlas of genetic correlations across human diseases and traits. Nature Genetics, 47(11):1236–1241, 2015. URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC4797329/. ↩

  2. Sara L Pulit, Charli Stoneman, Andrew P Morris, Andrew R Wood, Craig A Glastonbury, Jessica Tyrrell, Loïc Yengo, Teresa Ferreira, Eirini Marouli, Yingjie Ji, and others. Meta-analysis of genome-wide association studies for body fat distribution in 694 649 individuals of european ancestry. Human molecular genetics, 28(1):166–174, 2019. URL: https://academic.oup.com/hmg/article/28/1/166/5098227. ↩

  3. Benjamin B Sun, Joshua Chiou, Matthew Traylor, Christian Benner, Yi-Hsiang Hsu, Tom G Richardson, Praveen Surendran, Anubha Mahajan, Chloe Robins, Steven G Vasquez-Grinnell, and others. Plasma proteomic associations with genetics and health in the UK Biobank. Nature, 622(7982):329–338, 2023. URL: https://www.nature.com/articles/s41586-023-06592-6. ↩

  4. Giles Yeo. Gene eating: the story of human appetite. Orion Spring, 2018. URL: https://a.co/d/0bxNrVJZ. ↩

  5. Fernando Pires Hartwig, Neil Martin Davies, Gibran Hemani, and George Davey Smith. Two-sample mendelian randomization: avoiding the downsides of a powerful, widely applicable but potentially fallible technique. 2016. URL: https://academic.oup.com/ije/article/45/6/1717/3072174. ↩

  6. Luke J O’Connor and Alkes L Price. Distinguishing genetic correlation from causation across 52 diseases and complex traits. Nature Genetics, 50(12):1728–1734, 2018. URL: https://www.nature.com/articles/s41588-018-0255-0. ↩