Machine Learning for Human Genetics

Module: CIT423019

Credit: 6 ECTS

Moodle: https://www.moodle.tum.de/user/index.php?id=125481

Lecturers: Matthias Heinig, Julien Gagneur, Gabi Kastenmueller, Paolo Casale

Lecture: Thursdays, 14:00 - 15:30, Seminar room Taurus 1, Galileo (8120.EG.002)

Exercise: Tuesdays, 14:00 - 15:30, Seminar room 03.13.010

Start: Thur 15.10.2026

Lecture Language: English

Prerequisite:

• A course in Probability and Statistics. To check your level, go through before the lecture our primer in statistics, probability, and the multivariate Gaussian, i.e., make sure you are able to do the exercises available here: https://drive.google.com/drive/folders/1wcl04BtwjGJuC70IKT62hg5MBS0ilF_m
  
• Basics in biology and genetics (refresh your high-school level). 
  A refresher for the basic biology concepts required in this lecture can be found in Prof. Jonathan Pritchard’s online book: https://web.stanford.edu/group/pritchardlab/HGbook.html
  • Chapter 1.2 (A genome owner’s starter pack) to cover the basics.
  • Chapter 1.3 (Human genome variation and why it matters).
• Recommended: An introductory course in data science

Content:

This is a two-part module: (1) Seven lectures introduce the basics of systems genetics, and statistical models employed. Following the lecture, the hands-on tutorial empowers students to write code that directly solves that week's driving question. In addition, we will provide a script covering the statistical concepts and an online book covering the fundamentals of biology and genetics required for the lectures. We expect the students to be familiar with these before the start of the lectures. 

This is followed by (2) an eight-week hands-on project, which will be suggested and supervised by lab members from the lecturers. This way, you will have a unique opportunity to work on ongoing research projects in the respective lab.

During the lectures, the following topics will be covered:

• Which factors shape human phenotypes?
  Introduction and fundamentals of human genetics
• What is your genetic risk?
  GWAS, Polygenic Risk Scores (PRS)
• What are the underlying genes? (Target Discovery)
  Fine-mapping and gene-mapping
• How do genetic variations affect molecular processes?
  eQTL mapping, colocalization, the omnigenic model
• Which molecular processes causally affect phenotypes?
  Metabolomics, causal inference, Mendelian Randomization
• How can we generalize to unseen variants?
  Variant effect prediction, Rare variant association testing
• Which variants cause diseases?
  Rare disease diagnostics, multi-omics (RNA-seq, proteomics) for diagnostics

 

Intended Learning Outcomes: At the end of the module, students understand and are able to practically implement:

• Basic concepts in human genetics
• Challenges of complex trait genetics, properties of genetic data, and the role of rare vs. common variants
• GWAS and the computation of Polygenic Risk Scores (PRS)
• Methods for adjustment for multiple testing, including permutation-based FWER correction
• Fine-mapping methods to identify causal variants and underlying genes
• Gene expression analysis and mapping of eQTLs
• Causal inference methods, such as Mendelian randomization
• Variant effect predictors
• Rare variant association testing
• Omics approaches for rare disease diagnostics
• Apply the above techniques to real problems in systems genetics. Evaluate model performance, calibration, and provide biological interpretation of its application to real data.