BaNDyT: Bayesian Network modeling of molecular Dynamics Trajectories 1. BaNDyT introduces a Bayesian Network (BN) approach to molecular dynamics (MD) simulation data, offering a powerful tool to uncover dynamic dependencies between protein residues. Unlike traditional proximity-based models, BaNDyT identifies both local and allosteric interactions, bringing a new perspective on protein function. 2. The software enables fully data-driven insights into molecular dynamics, moving beyond user-based bias. BaNDyT leverages Bayesian networks to interpret MD trajectories, making it a unique resource for researchers analyzing complex protein systems like G protein-coupled receptors (GPCRs). 3. A significant feature of BaNDyT is its dual capability to analyze single residues and inter-residue pairs. This enables in-depth modeling of protein-protein interfaces, key for understanding interaction dynamics in GPCR:G protein complexes. 4. BaNDyT provides scalability for large systems and long timescale simulations, supporting researchers with MD trajectories across varied biomolecular structures, from small proteins to polymeric materials. 5. The software offers an interpretable, unsupervised machine learning model, with a Python interface that allows researchers to fine-tune MD data analysis and visualize networks using Cytoscape, facilitating comprehensive protein interaction studies. 6. Key advantages include the capacity to infer both direct and indirect dependencies within proteins, allowing insights into functional residues that play roles in allosteric regulation and stability, potentially informing targeted therapeutic design. 7. The BN output reveals critical nodes and network properties such as weighted degree, helping researchers identify high-impact residues and interactions that influence protein function and allosteric communication. 8. By modeling GPCR interactions with G proteins, BaNDyT has revealed new insights into selective coupling mechanisms, showing promise for wider applications in protein complex analysis and drug discovery. @emukhaleva @Vaidehilab 💻Code: github.com/bandyt-group/band… 📜Paper: biorxiv.org/content/10.1101/… #BayesianNetwork #MolecularDynamics #ProteinInteractions #GPCR #MachineLearning #ComputationalBiology