Mendelian Randomization

An online short course

This course aims to provide an introduction to the conduct, assumptions, strengths and limitations of Mendelian randomization, including the use of up-to-date methods for sensitivity analyses that explore likely violation of Mendelian randomization assumptions.

Course date Due to demand this standalone course will run twice:
17 - 19 March 2021
21 - 23 April 2021
Course fee £660
Course Organisers Professor Debbie Lawlor & Dr Kaitlin Wade

Prerequisites

Please ensure you meet the following prerequisites before booking:

Knowledge Prior experience of using Mendelian randomization is not required, but participants should have an understanding of aetiological epidemiological principles, and ideally be working on causal population health questions.
Those intending to take this course should already understand epidemiological principles and have knowledge and skills in statistical analysis to the level of running, and correctly interpreting results from, multivariable regression analyses. 
They must have experience in running such analyses efficiently in Stata and/or R as all practicals on the course will be offered in both Stata and R and the focus of these practicals will be on Mendelian randomization (not learning how to use the statistical packages). 
Note: it is not necessary for those participating in the course to be able to use both Stata and R, but you must be able to use one of these.
Software Participants who would like to use Stata need to have installed Stata version 16* in advance of the course.
Those who would like to use R (latest version) need to have this downloaded onto their computer.
The operating system can be either Windows or Mac and participants should consider having a fast internet speed for synchronous/live sessions and for downloading data and scripts.
*Internal University of Bristol participants will be provided with access to Stata version 16 on the first day of the course.

Introduction

Mendelian randomization is a study that uses genetic variants as instrumental variables to test the causal effect of a (non-genetic) risk factor on a disease or health-related outcome. Since its first proposal in 2003, it has been increasingly used to determine population causal effects using observational data. It is used in a large amount of the applied research in the MRC Integrative Epidemiology Unit (IEU) and throughout Population Health Sciences at the University of Bristol Medical School. Academics working in the IEU and the University of Bristol (including those who are tutors on this course) have been at the forefront of developing methods for assessing and limiting potential biases with this approach. 

Course format

This 3-day course will be online and consist of learning activities set by the tutor including lectures (synchronous and asynchronous), small group work, discussions, individual tasks, and computer practical activities. Directed and self-directed learning will include activities such as reading, accessing web-based supplementary materials, critical appraisal and completion of quizzes. All teaching will be conducted online using Blackboard.

Course objectives

By the end of the course, students will be able to:

  1. describe the principles and assumptions of instrumental variable analyses;
  2. discuss the properties of genetic variants that make them suitable to be used as instrumental variables;
  3. explain the strengths and limitations of one-sample and two-sample Mendelian randomization for addressing population health causal questions;
  4. conduct a (straightforward) one-sample and two-sample Mendelian randomization analysis;
  5. describe the concepts behind sensitivity analyses to test for potential violation of the key assumptions of Mendelian randomization;
  6. apply up-to-date sensitivity analyses in one- and two-sample Mendelian randomization analyses;
  7. critically appraise Mendelian randomization papers and analyses;
  8. identify the key features required in writing reproducible and transparent Mendelian randomization papers;
  9. design a Mendelian randomization study of an exposure/outcome pair.

Who the course is intended for

The course is intended for anyone who wants to be able to undertake Mendelian randomization analyses. It is an introductory to intermediate course. The course will not include any genetic epidemiology teaching, nor how to undertake a genome-wide association study. However, genetic epidemiology and the ability to complete a genome-wide association study are NOT a prerequisite for being able to understand this course. The course will not cover more complex Mendelian randomization methods such as two-step or network Mendelian randomization.

Course outline

Students will learn about one-sample and two-sample Mendelian randomization, including their assumptions and gaining practical experience of how to apply these methods to real data. They will also learn about a range of sensitivity analyses that explore likely violation of the assumptions of Mendelian randomization. Over the duration of the course, there will be opportunities to ask questions about Mendelian randomization to experienced tutors working in the field, work in groups to design a Mendelian randomization study on a particular topic, undertake a critical appraisal of a Mendelian randomization paper and consider the key aspects of writing a manuscript including Mendelian randomization analyses.
 
The course will take place over three days.
 
The first day will include a recap of genetic and epidemiological concepts useful for conducting, understanding and interpreting Mendelian randomization analyses; an introduction and computer practical on one-sample Mendelian randomization and the principals of two-sample Mendelian randomization.
 
Day two will expand on the two-sample Mendelian randomization principals with computer practicals focusing on data harmonization, analyses and interpretation of results. Participants will also be introduced to the MR-Base platform and how to use it appropriately; explore recent advances in and future directions of Mendelian randomization and the use of Mendelian randomization in drug discovery and target validation.
 
In the final day, students will undertake a critical appraisal of a Mendelian randomization paper, as well as learn how to write and design a Mendelian randomization study of their own. We will also contextualise Mendelian randomization in the broader field of epidemiology by introducing the importance of triangulation.
 
Recommended reading

Davey Smith G, Hemani G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet. 2014;23(R1):R89-R98. doi:10.1093/hmg/ddu328.

Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International Journal of Epidemiology 2015; 512–525. doi: 10.1093/ije/dyv080.

Lawlor DA. Two-sample Mendelian randomization: opportunities and challenges. International Journal of Epidemiology 2016; 908–915. doi: 10.1093/ije/dyw127.

Haycock PC, Burgess S, Wade KH, Bowden J, Relton C, Davey Smith. Best (but oft-forgotten) practices: the design, analysis, and interpretation of Mendelian randomization studies. American Journal of Clinical Nutrition 2016; 103: 965–78.

Online Course Bookings


Bookings are open for online courses running in 2021.

I enjoyed all lectures and practicals led by the main tutors. They are all very knowledgeable and passionate about the topic, and created the warm and motivating atmosphere during the course.

Course feedback, February 2019

A great introduction, particularly highlighting the importance of sensitivity analysis and common pit falls in Mendelian Randomization.

Course feedback, July 2019

Coronavirus (COVID-19)

We may need to make responsive changes to our courses at short notice in order to follow the latest Public Health, Government and University guidance on coronavirus (COVID-19).

Edit this page