Abstract:Side chain flexibility is an important factor in ligand binding. In order to determine the extent to which side chain flexibility is involved in ligand binding, a knowledge-based approach was taken. A database composed of examples of protein structures in the presence or absence of a given ligand is used to analyze which side chains undergo side chain conformational changes. Such an analysis has determined that up to 40% of binding site do not present side chain conformational changes. A total of three residues undergoing side chain conformational changes encompass approximately 85% of the binding sites studied. When analyzing the propensities of different amino acids to undergo side chain conformational changes we find that there are considerable differences between different amino acids. A support vector machine learning approach was used to create a classifier system utilizing information about the solvent accessible area as well as flexibility scale value of each specific side chain to be predicted together with its neighboring side chains. An accuracy level of 70% is reached using this approach. The fact that a small number of residues undergo side chain conformational changes in the majority of binding sites makes it feasible to introduce side chain flexibility in docking simulations. An algorithm has been developed for introducing side chain flexibility utilizing a hybrid genetic-algorithm/exhaustive-search procedure and a surface complementarity based scoring function. This approach is implemented in the software tool FlexAID. FlexAID utilizes a rotamer library to create alternative conformations for a list of residues that are exhaustively searched during the docking simulation. The performance of FlexAID in rigid local as well as global simulations falls in the 70-80% range for both local and global simulations.

Deep Learning Model for Efficient Protein–Ligand Docking with Implicit Side-Chain Flexibility

Graph-convolution neural network-based flexible docking utilizing coarse-grained distance matrix

Equivariant Flexible Modeling of the Protein-Ligand Binding Pose with Geometric Deep Learning.

ApoDock: Ligand-Conditioned Sidechain Packing for Flexible Molecular Docking

FlexVDW: A machine learning approach to account for protein flexibility in ligand docking

Deep Learning for Protein-Ligand Docking: Are We There Yet?

Protein flexibility upon ligand binding: Docking predictions and statistical analysis

Re-Dock: Towards Flexible and Realistic Molecular Docking with Diffusion Bridge

Harnessing Deep Learning for Enhanced Ligand Docking.

Comprehensive Assessment of Flexible-Ligand Docking Algorithms: Current Effectiveness and Challenges

A fully differentiable ligand pose optimization framework guided by deep learning and traditional scoring functions

A Protein-Peptide Docking Program With Modeling Receptor Flexible Areas

A fully differentiable ligand pose optimization framework guided by deep learning and a traditional scoring function

DeltaDock: A Unified Framework for Accurate, Efficient, and Physically Reliable Molecular Docking

DeepDock: Enhancing Ligand-protein Interaction Prediction by a Combination of Ligand and Structure Information

DEELIG: A Deep Learning Approach to Predict Protein-Ligand Binding Affinity

Combining Docking Pose Rank and Structure with Deep Learning Improves Protein–Ligand Binding Mode Prediction over a Baseline Docking Approach

DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking

Exploration on Learning Molecular Docking with Deep Learning Models

DLIGAND2: an Improved Knowledge-Based Energy Function for Protein–ligand Interactions Using the Distance-Scaled, Finite, Ideal-Gas Reference State

A new paradigm for applying deep learning to protein–ligand interaction prediction