Datasets used in the benchmarking study of MR methods

We conducted a benchmarking analysis of 16 summary-level data-based MR methods for causal inference with five real-world genetic datasets, focusing on three key aspects: type I error control, the accuracy of causal effect estimates, replicability, and power.

The datasets used in the MR benchmarking study can be downloaded here:

1. "dataset-GWASATLAS-negativecontrol.zip":  the GWASATLAS dataset for evaluation of type I error control in confounding scenario (a): Population stratification
2. "dataset-NealeLab-negativecontrol.zip": the Neale Lab dataset for evaluation of type I error control in confounding scenario (a): Population stratification;
3. "dataset-PanUKBB-negativecontrol.zip": the Pan UKBB dataset for evaluation of type I error control in confounding scenario (a): Population stratification;
4. "dataset-Pleiotropy-negativecontrol": the dataset  used for evaluation of type I error control in confounding scenario (b): Pleiotropy;
5. "dataset-familylevelconf-negativecontrol.zip": the dataset used for evaluation of type I error control in confounding scenario (c): Family-level confounders;
6. "dataset_ukb-ukb.zip": the dataset used for evaluation of the accuracy of causal effect estimates;
7. "dataset-LDL-CAD_clumped.zip": the dataset used for evaluation of replicability and power;

Each of the datasets contains the following files:

1.  "Tested Trait pairs": the exposure-outcome trait pairs to be analyzed;
2. "MRdat" refers to the summary statistics after performing IV selection (p-value < 5e-05) and PLINK LD clumping with a clumping window size of 1000kb and an r^2 threshold of 0.001.
3. "bg_paras" are the estimated background parameters "Omega" and "C" which will be used for MR estimation in MR-APSS.

Note:

1. The formatted dataset after quality control can be accessible at our GitHub website (https://github.com/YangLabHKUST/MRbenchmarking).
2. The details on quality control of GWAS summary statistics, formatting GWASs, and LD clumping for IV selection can be found on the MR-APSS software tutorial on the MR-APSS  website (https://github.com/YangLabHKUST/MR-APSS).
3. R code for running MR methods is also available at https://github.com/YangLabHKUST/MRbenchmarking.