open Menu
Search icon

SP 1

Identification of risk alleles and risk profiles

Over the last 5 years, we participated in global consortia which elucidated almost all currently known risk loci of coronary artery disease and stroke. Such leading role became possible after pioneering work on individual genome-wide association studies (GWAS) carried out by the group. However, there is ample evidence for „missing heritability“, which may result from currently not considered rare variants or gene-gene or gene-environment interactions.
Within this project, we will use all available data in a pipeline for systematic analysis of rare variants and interaction models. In co-operation with other subprojects, we thus aim at explaining further large proportions of the biology and functionality of the investigated diseases, thus defining possible therapeutic targets. For this, we combine our methodological and clinical expertise with large, well phenotyped data. As a result of our analyses new candidates – single variants or profiles – for functional and bio-informatics follow-up in other subprojects are generated. Furthermore, these will be utilized in prospective data to investigate their value as predictive entities.

Work package 1: Previous GWAS have focused on univariate analyses of single autosomal SNPs mostly using additive genetic models. We will further analyze the data by i) 1000G imputation , ii) investigating specific genetic models, and iii) X chromosomal loci.
Work package 2: Previous genotyping chips have focused on common variants. We now analyze exome-wide data generated by typing >25,000 cases and controls on the ExomeArray as well as by whole exome-sequencing in unrelated cases and in families.
Work package 3: We will investigate the role of gene-gene interactions with repsect to these phenotypes.
Work package 4: Given established clinical risk factors for CAD and stroke, systematic interaction analyses on a GWA level are promising and performed as part of this package. Further, gene-environment dependence will be analyzed using machine learning techniques such as random forest and multifactor dimensionality reduction.