Analyses of X-chromosomal SNPs and kidney traits
Last Updated: 29/09/2023
Supporting code for the following draft:
- Working title: X-chromosome and kidney function: Evidence from a multi-trait genetic analysis of 908,697 individuals reveals sex-specific and sex-differential findings in genes regulated by androgen-response elements
- Short title: CKDGen ChrX
- Analysis Team / Repository Contributor: Markus Scholz, Katrin Horn, Andreas Kühnapfel and Janne Pott
- Writing Team: Markus Scholz, Katrin Horn, Afshin Parsa, Anna Köttgen, Pascal Schlosser, Cristian Pattaro
We performed a trans-ethnic X-chromosome-wide association study of 7 kidney traits (estimated glomerular filtration rate (eGFR), serum uric acid (UA), urine albumin-creatinine ratio (UACR), blood urea nitrogen (BUN), chronic kidney disease (CKD), gout and microalbuminuria (MA)). Sex-stratified and combined analyses were performed in up to 40 studies including up to 908,697 subjects considering up to 1,032,701 SNPs. Genome-wide significant loci were tested for sex-interactions and were compared between traits. A number of secondary analyses were performed to allocate candidate genes to the discovered loci.
For more information, please contact Markus Scholz (markus.scholz@imise.uni-leipzig.de)
If you want to reproduce our results, you will need to customize a source file, indicating
- path to R library (please use R Version 4.x, all necessary packages are listed in the source file)
- path to data (summary statistics will be available on zenodo)
- path to PLINK2
- path to GCTA
- path to GTEx v8 data
- path to NEPTUNE
- path to HUNT data
- path to UKBB data
- Locus definition: index SNP +/- 500 kb; merging of overlapping loci (union of loci)
- Interaction analysis: difference test of the stratified associations, correcting for beta-beta correlation between males and females
- Co-localization analyses of male and female signals: test of the stratfied associations for shared causal signal
- Relevance to kidney function and CKD: Comparison of eGFR and UA index SNP effects with those in CKD, BUN, and gout (one-sided test)
- Identification of independent variants: GCTA - conditional joint (COJO) analyses
- Reference data set: UKBB (best-guessed genotypes)
- COJO select: step-wise forward selection to indentify independent variants
- COJO conditional: Estimation of association statistics conditional to previously selected variants (in case of multiple independent variants at a locus)
- LD estimation: LD between independent variants per locus using PLINK2
- Credible set analyses: Calculation of Approximate Bayes Factors using (conditional) effect estimates and standard errors
- Co-localization analyses with eQTLs: test of the associations for shared causal signal with gene-expression of nearby genes using GTEx v8 and NEPTUNE data
- Extraction of eQTL data of nearby genes in 51 tissues
- Colocalization analyses per phenotype, tissue and gene
- Checks and plots
- Analysis of overlap of eGFR and UA signals:
- LD estimation: LD between eGFR and UA index SNPs using PLINK2
- Co-localization analyses of eGFR and UA: test of the associations of eGFR and UA for shared causal signal
- Replication analysis: replicate findings of eGFR in HUNT study (one-sided)
- X-chromosomal heritability: GCTA - restricted maximum likelihood (REML) analyses
- Reference data set: UKBB (genotyped SNPs)
- REML: estimation of heritability for eGFR and UA
- REML bivariate: estimation of genetic correlation between eGFR and UA
- Lookup of reported variants: replicate findings of Graham et al, Kanai et al and Sakaue et al (successful if nominal significance and concordant effect is observed)
- Trans-ethnic meta-regression analysis: accounting for mixed ethnicities by using MR-Mega
- Get an overview of MR-Mega results and check results for MetaGWAS loci
- Generate Forest plots for MR-Mega hits
- Study quality control and harmonization: pipeline from GenStatLeipzig, not yet available on GitHub
- Meta-analyses: pipeline from GenStatLeipzig, not yet available on GitHub
- Bioinformatic annotation: pipeline from GenStatLeipzig, not yet available on GitHub
- Lookup of reported variants: manual evaluation of GWAS catalog data of recent publications from Graham et al, Kanai et al and Sakaue et al
- Trans-ethnic meta-regression analysis: done by Alexander Teumers lab (Greifswald, Germany), not yet shared
- Assignment of candidate genes: manual evaluation of all previous applied methods, no code available
- Miami-Plot of variants associated with eGFR and UA:
- color coding: grey, overall; blue, males; red, females; black, not genome-wide significant
- novelty coding: candidate gene in box novel, candidate gene without box kown
- sex interaction coding: bold italic gene names indicate loci with sex interactions (--> see script F1_MiamiPlot.R)
- Regional Association Plot of locus 7 at Xq21.1:
- panel with all, male, female
- Beta-Beta-Plot of SNP-by-Sex interaction analysis:
- panel with results for eGFR and UA (--> see script 02_sex_ia.R)
- Heatmap of cross-phenotype comparison of eGFR and UA loci:
- six phenotypes: eGFR, UA, BUN, CKD, UACR and MA
- 23 SNPs: index SNP per locus + female-specific index SNP of region 7 (--> see script F2_Heatmap_phenotypes.R)
- Results of eQTL-colocalization analysis:
- six phenotypes: eGFR ALL, MALE, FEMALE and UA ALL, MALE, FEMALE
- eight genes: CDKL5, CDK16, USP11, ARMCX2, TCEAL3, MORF4L2, ACSL4, SLC25A5 (--> see scripts starting with 07_)
- RA-Plots of all loci for ALL, MALE, FEMALE
- Bar-Plot of X-chromosomal heritability of eGFR and UA (--> see script 10_b_Heritability.R)
- Beta-Beta-Plot of validation of eGFR hits in HUNT study (--> see script 09_replication_HUNT.R)
- Forest Plots for SNP rs4328011 (--> see script 12_b_MR-Mega_make_forest_plots.R)
- Forest Plots of MR-MEGA findings (--> see script 12_b_MR-Mega_make_forest_plots.R)
- Credible sets and missense mutations (--> see script F4_MainFigure4_PostProbByCredSetSize.R)
- Study design (not done with a script)
- Beta-Beta-Plot of index SNPs comparing trans-ethnic metaGWAS and Europeans only metaGWAS
- QQ-Plots of phenotypes and subgroups
- Results of the transethnic X chromosome-wide association analysis of eGFR and UA (--> see script T1_MainTable1.R)
- Analysis of the overlapping of eGFR and UA loci (--> see script T2_MainTable2.R)
--> see script ST_SupplementTables.R
- Description of participating studies: study design and phenotype distribution (as received from participating studies)
- Genotyping and imputation information of participating studies (as received from participating studies)
- Number of data sets, samples and SNPs contributing to the different association analyses (--> see script ST_SupplementTables.R)
- Comparisons between sexes, interaction and co-localization (--> see scripts 02_sex_ia.R and 03_coloc_sex_ia.R)
- Cross-phenotype comparision (--> see script 04_lookup_TopHits_otherTraits.R)
- Independent variants per locus and analysis group (--> see script 05_b_Cojo_Select.R and 05_c_Cojo_Conditional.R)
- Annotation of 99% credible sets (done with pipeline from GenStatLeipzig)
- Co-localization of genetic association signals and eQTLs (--> see scripts starting with 07_)
- Validation of eGFR associations in HUNT study (--> see script 09_replication_HUNT.R)
- Look-up of SNPs previously reported for UA, eGFR, creatinine and BUN (reported variants of three GWAS of kidney traits: Graham et al, Kanai et al and Sakaue et al) (--> see script 11_Lookup_Candidates_SNPs.R)
- Results of meta-regression analysis of the 23 index SNPs (--> see scripts 12_a_MR-Mega_results_check.R)
- Additional genome-wide significant associations due to meta-regression (--> see script 12_a_MR-Meta_results_check.R)
- Look-up of sex-biased gene-expressions of candidate genes assigned to genetic sex-interactions (--> see script 13_Lookup_sexbiasedGE_eQTLs_PAR_ARE.R)
- Co-localization of genetic association signals of CKD traits and testosterone (--> see script 07_d_coloc_testo.R)
- Look-up of eQTLs in case of positive co-localization (--> see script 13_Lookup_eQTLs_leadSNPs.R)