New research from Faidon Brotzakis and co-workers sheds light on targeting the structural dynamics of the partially disordered protein AR-V7 using a deep ensemble docking pipeline.
AR-V7, a splicing variant associated with prostate cancer, presents significant challenges for traditional in-silico drug discovery due to its highly dynamic conformational landscape and numerous transient binding sites. In this study, the authors introduce a computational approach that reduces the dimensionality of binding site space by 90-fold and integrates machine learning with physics-based molecular docking—enhancing the identification of multisite small-molecule binders by 17-fold. Among the top candidates, ChEMBL22003 modulates AR-V7’s conformational ensemble by decreasing entropy and altering solvent exposure at key functional regions. These findings highlight its potential as a phase separation modulator, paving the way for new therapeutic strategies targeting intrinsically disordered proteins involved in cancer.
To read the full article, visit: https://pubs.acs.org/doi/10.1021/acs.jctc.5c00171