Practical 4 - Mendelian randomization

Practical 4 – Mendelian randomization


Q1: How is two-sample MR different from one-sample MR?

One sample = SNP, exposure and outcome are all measured in the same individuals or study
populations
- Bias in the direction of the confounded observational association (any direction)

Two sample = SNP-exposure association (G-X) and SNP-outcome association (G-Y) estimated in two
independent samples




 Exposure and outcome do not need to be measured in the same sample
 Non-overlapping samples
 Regression coefficients can be used for calculating Wald ratios
 Massive amount of GWAS data publicly available with which two-sample MR can be
performed
 Weak instruments = bias towards the null (predictable)

Q2: What is meant by summary statistics? What types of variables are in this database? (Think of
individual-level data for comparison.)

Publicly available GWAS data are summary data. Summary statistics are defined as the aggregate p-
values and association data for every variant analysed in a genome-wide association study (GWAS).

There is no individual data, only data on study level.

Types of variables:
- Measures of association:
o OR
o Beta
- P-values
- 95% CI
- Gene region
- SNP-risk allele
o Frequency