The de.NBI & ELIXIR-DE and GHGA Knowledge Series: variant prediction with dms data
- 03 Mar 2026
Deep mutational scanning (DMS) experiments provide powerful, high-resolution insights into the functional impact of genetic variants, but real-world datasets are often incomplete, heterogeneous, and difficult to interpret or reuse. This poses challenges for researchers who want to leverage DMS data for variant interpretation and medical research applications.
This webinar provides a practical, end-to-end walkthrough of accessing, interpreting, and extending DMS data using open community resources and machine learning models. Guidance is provided on how these approaches can be applied responsibly in both research and translational contexts.
Starting with an introduction to DMS and its relevance for biomedical research, attendees will learn how to integrate DMS data from MaveDB, the primary repository for variant effect data, into bioinformatics workflows. The core of the talk will focus on imputation strategies for incomplete DMS datasets, including design considerations, limitations, and best practices. Using the VEFill model as a concrete example, this session illustrates how sequence embeddings, evolutionary information, and physicochemical features are combined to predict missing variant effects and support diverse downstream applications.
This webinar is aimed at students, early career researchers, clinicians, bioinformaticians and everyone interested in advancing their understanding of functional genomics through open-source bioinformatics workflows and predictive protein modelling.
The webinar will be held by Polina Polunina from the University of Freiburg.
The event is in English, it takes approx. 60 minutes and will be held on the 3rd March 2026 at 13:00 CET (available afterwards on demand). Attendance is free, but we ask for registration.