Presentation

Professor and e-Science Faculty Björn Wallner, focusing on protein structure prediction, modelling and dynamics using AI and machine learning. He was the first to develop the multiple sequence sequence embeddings for deep learning applications, which now form the basis for AlphaFold. He has been deeply involved in developing the Rosetta software suite for modelling macromolecules. He successfully integrated sparse proximity constraints to achieve unprecedented accuracy, enabling molecular dynamics studies of proteins without known structure.  He has used computational modelling combined with experimental techniques to study mutation-dependent conformational ensembles in the P. Aeruginosa gene regulator MexR, responsible for antibiotic resistance. By combining explicit modelling of atomic interactions with ambiguous distance constraints for the first time he described the interaction between the disorded Myc protein to Pin1 at the molecular level. In parallel he developed a predictor of local coordinate error of protein structure models that was ranked no 1 in the community-wide benchmark CASPs. The local error estimates have been incorporated into the Phaser software for molecular replacement making it possible to now solve the crystallographic phase problem for more targets using less accurate models. He has predicted protein-protein interactions combining structural comparison and modeling with sequence data and AI to pinpoint evolutionary events responsible for interactions.  Recent publications in Bioinformatics, Structure and Nature Structural and Molecular Biology.

Publications Show/Hide content

2022

2021

2020

News Show/Hide content

Publications Show/Hide content

2022

2021