Workshop

Understanding ML-based protein structure prediction

In the second workshop we will take an indepth look at some of the leading ML-based structure prediction models.

Following a brief introduction/refresher of protein basics, we will look at the way different inputs are taken by the model to produce a refined representation of the information extracted from the inputs.

We will then dive into the actual prediction model looking at the mechanics/architecture used in different models. Finally we will take a comparative overview of different models to better understand the challenges ahead. 

Please bring your laptop

Tutor:

Amijai Saragovi - is a postdoctoral researcher in the Baker Lab, where he combines AI-powered protein design with experimental innovation to create functional protein-based nanomaterials. His work focuses on harnessing proteins to template semiconductor growth, offering a scalable and sustainable alternative for next-generation electronics. As he transitions to establish his research in Lund University, Saragovi envisions a future where biofabricated materials redefine energy-efficient computing, photonics, and quantum technologies—pushing biology beyond evolution to engineer entirely new functionalities.

Agenda:

17:30 – 17:45 – Meet & Greet
17:45 – 18:05 – The basics - introduction/refresher to proteins
18:05 – 18:35 – Data inputs in AlphaFold2/ RoseTTA fold: MSA, templates and pairing 
18:35 – 19:00 – Meet & Eat
19:00 – 19:20 – Processing and refining: refining the input for prediction
19:20 – 19:50 – Predicting protein structures: mechanics and output
19:50 – 20:30 – A comparative look at existing prediction models and the challenges ahead (a practice session) 

sponsored by:

Capish