#39 · Structural bio & protein design

Structural modelling links disease-associated variants to altered protein features

The common thread is structure-based inference applied to disease-associated sequence variation. A preprint predicts condensation-related signatures localized to two C-terminal regions of the cervid prion protein across PRNP variants associated with chronic wasting disease. Another applies AI-based modelling to 109,282 human protein interaction structures and maps pathogenic variants onto predicted interfaces. A third uses an AI protein co-scientist workflow, combining Boltz-2, rDock, AlphaFold3 and HDOCK, to simulate how OGT TPR deletions and X-linked intellectual disability substitutions alter structure. The published study links conserved water molecule sites in PDB structures to enrichment for pathogenic missense variants, with MD simulations of GCase tying disruption of these sites to pathogenicity. What is new is the extension of AI-assisted modelling to disease-variant mutants and the identification of conserved hydration sites as a structural correlate of missense pathogenicity.