Correlation Studies of Intrinsic Disordered Protein

Within proteins the residues form highly correlated networks through hydrogen bonds and hydrophobic interactions, enabling proteins to to effectively resist external environmental influences and maintain stable structure. And the folding processes of many proteins can be interpreted as first-order transition. On the other hand, the understanding about the possible correlation between various degrees of freedom within intrinsically disordered protein is very limited. We used molecular dynamics simulation, statistical mechanics theory and developed deep learning algorithm to study the aforementioned systems. We identified the significant spatial correlation between DOPA and positively charged residues in the intrinsically disordered mussel adhesive protein, providing the protein with strong adsorption capabilities. We revealed the correlation between the dynamic interaction between RGG domains and their conformation fluctuation. And we developed GraphSAGE-embedded LSTM network to capture the dynamic nature of IDP-involved interaction and predict their behavior.