Design and quality control of large-scale two-sample Mendelian randomisation studies

Abstract

Background: Mendelian randomization studies are susceptible to meta-data errors (e.g. incorrect specification of the effect allele column) and other analytical issues that can introduce substantial bias into analyses. We developed a quality control pipeline for the Fatty Acids in Cancer Mendelian Randomization Collaboration (FAMRC) that can be used to identify and correct for such errors. Methods: We invited cancer GWAS to share summary association statistics with the FAMRC and subjected the collated data to a comprehensive QC pipeline. We identified meta data errors through comparison of study-specific statistics to external reference datasets (the NHGRI-EBI GWAS catalog and 1000 genome super populations) and other analytical issues through comparison of reported to expected genetic effect sizes. Comparisons were based on three sets of genetic variants: 1) GWAS hits for fatty acids, 2) GWAS hits for cancer and 3) a 1000 genomes reference set. Results: We collated summary data from six fatty acid and 49 cancer GWAS. Meta data errors and analytical issues with the potential to introduce substantial bias were identified in seven studies (13%). After resolving analytical issues and excluding unreliable data, we created a dataset of 219,842 genetic associations with 87 cancer types. Conclusion: In this large MR collaboration, 13% of included studies were affected by a substantial meta data error or other analytical issue. By increasing the integrity of collated summary data prior to their analysis, our protocol can be used to increase the reliability of post-GWAS analyses. Our pipeline is available to other researchers via the CheckSumStats package (https://github.com/MRCIEU/CheckSumStats).