Title:	AI-based Structural Analysis of Glycosylation Machinery in Bacteroidota
DNr:	Berzelius-2025-284
Project Type:	LiU Berzelius
Principal Investigator:	Andre Mateus <andre.mateus@umu.se>
Affiliation:	Umeå universitet
Duration:	2025-08-31 – 2026-03-01
Classification:	10601
Homepage:	https://mateuslab.com/
Keywords:

Abstract

Protein glycosylation in the bacterial phylum Bacteroidota is diverse and functionally important, but the structural basis of this process is poorly understood. Glycosylation influences microbial fitness and host–microbe interactions, yet the enzymes and transfer systems that mediate these modifications remain largely uncharacterized. This project will use AlphaFold3 to generate structural predictions of glycosylation machinery in Bacteroidota, with a focus on: 1. Glycosyltransferases (GTs) and their matching with activated monosaccharide donors. 2. Oligosaccharyltransferase (OTase) components, including analysis of acceptor peptide binding and donor recognition. 3. Comparative evaluation across species to detect conserved and divergent structural features. Predictions will be combined with AI/ML clustering to identify structural motifs and putative functional groupings. The work will provide proof-of-concept for large-scale exploration of glycosylation systems using ML-based structural biology, and generate hypotheses on how glycosylation shapes Bacteroidota diversity. This pilot will establish workflows for AlphaFold3 on Berzelius, test integration with downstream ML analyses, and generate a small but valuable set of glycosylation protein models. Results will inform future larger-scale studies, while already delivering new insights into bacterial glycobiology and microbiome–host interactions.