Prediction of peptide properties for proteomics
Course outline and goals
This course focuses on the application of machine learning to predict different peptide properties (regression or classifications tasks) that are relevant to mass spectrometry-based proteomics [1].
During this course, independent of the assigned prediction task, students are going to:
- perform literature research of a pre-defined topic
- get a general understanding of machine learning and how to apply machine learning to biological data [2,3]
- develop and correctly evaluate a machine learning model including parameter optimization [2,3]
- present milestones and final results in various presentations to the other students and supervisors
- summarize results in a paper-like scientific report at the end of the course
- gain a first look at the peer-review process by critically evaluating the reports of other students
Students will work in groups of 3-4. They will present their topic, background as well as their datasets together. They will work on the same prediction task, but will follow different approaches as discussed with their supervisor. In the end, each group will merge their result, present them and and final conclusion in a talk and a written scientific report.
Additionally, there is a challenge for each peptide property hosted on hugging face, where student can participate individually on voluntary basis. The submission of each peptide property with the best metric that is above a predefined baseline will receive a bonus.
Course summary
| Type | 5SWS (1 year module, 2 SWS/Summer, 3 SWS/Winter) |
|---|---|
| ECTS | 9 |
| Lecturer | Mathias Wilhelm, Zixuan Xiao, Cemil Can Saylan, Victor-George Giurcoiu, Joel Lapin |
| Time | Monday, 14:00 - 16:00 |
| Location | In person in Freising/Weihenstephan - Maximus-von-Imhof-Forum 3, room OG-L 19 (possible to join via Zoom on request) |
| Language | English |
Kick-off meeting
ALL TIMES LISTED ARE NOT FINAL YET AND ARE SUBJECT TO CHANGE!
The kick-off meeting is scheduled to take place in-person on Monday, 20.04, 14:00-16:00 in Freising, but may be moved in consultation with the participants. Attendance is mandatory.
All important information including detailed description of the topics will be made available via Moodle (link will be provided during kickoff).
Timeline
| Date | Type | Topic |
|---|---|---|
| 2026-04-20 | Lecture | Kickoff |
| 2026-04-27 | Lecture | Scientific presentations |
| 2026-05-04 | Lecture | Literature research / Topic selection |
| 2026-05-11 | Lecture | Introduction to machine learning |
| 2026-05-25 | Student presentation | Introduction talks |
| 2026-06-22 | Student presentation | First milestone talks |
| 2026-07-20 | Student presentation | Progress report + Q&A |
| 2026-10-12 | Lecture | Scientific writing and peer review process |
| 2026-10-19 | Student presentation | Second milestone talks |
| 2026-11-23 | Submission | Scientific report (draft) |
| 2026-12-07 | Submission | Peer-review revision |
| 2027-01-25 | Student presentation | Final presentation |
| 2027-02-01 | Submission | Scientific report (final) |
Topics
Each topic will be worked on by three students.
Prediction of collisional-cross-section
Prediction of retention time
Prediction of charge state
Prediction of fragment intensities
