Machine Learning for Regulatory Genomics

What it is about

In this module you will learn biological fundamentals of gene regulation and modern deep learning methods for modeling sequence-based regulatory processes. After completion you will be able to apply genome-wide experimental procedures and deep learning models for different stages of gene expression and biologically interpret the model predictions.

What you will be able to do

• Explain the main steps of gene expression from DNA accessibility to protein abundance
• Describe genome-wide experimental assays for different expression stages
• Explain the principle and applications of massively parallel reporter assays
• Apply deep-learning methods for sequence-based predictions
• Apply and interpret model interpretation techniques
• Use CNNs and transformer-based models on sequence data
• Evaluate the performance of deep-learning models for genome-wide assays
• Classify model predictions biologically and discuss limitations

What the module consists of

• LecturesConveying the biological basics, experimental methods and modeling approaches (six thematic lectures)
• Practical sessionsIn-class Python tutorials for applying the presented models
• Project workEight-week supervised project in a partner research laboratory with final report and presentation

Teaching method

• LecturesPresentation of the state-of-the-art in Regulatory Genomics and Vermittlung der Konzepte
• In-class tutorialsDirect application of concepts and models for deepening understanding
• Supervised projectHands-on experience with real biological/biomedical data and transfer of learned content to research topics