Abstract:background: computational molecular docking plays an important role in determining the precise receptor-ligand conformation, which becomes a powerful tool for drug discovery. In the past 30 years, most computational docking methods treat the receptor structure as a rigid body, although flexible docking often yields higher accuracy. The main disadvantage of flexible docking is its significantly higher computational cost. Due to the fact that different protein pock-et residues exhibit different degrees of flexibility, semi-flexible docking methods, balancing rigid docking and flexible docking, have demonstrated success in predicting highly accurate conformations with a relatively low computational cost. method: In our study, the number of flexible pocket residues was assessed by quantitative analysis, and a novel adaptive residue flexibility prediction method, named A-RFP, was proposed to improve the docking performance. Based on the homologous information, a joint strategy is used to predict the pocket residue flexibility by combining RMSD, the distance between the residue sidechain and the ligand, and the sidechain orientation. For each receptor-ligand pair, A-RFP provides a docking conformation with the optimal affinity. result: By analyzing the docking affinities of 3507 target-ligand pairs in 5 different values ranging from 0 to 10, we found there is a general trend that the larger number of flexible residues inevitably improves the docking results by using Autodock Vina. However, a certain number of counterexamples still exist. To validate the effectiveness of A-RFP, the experimental assessment was tested in a small-scale virtual screening on 5 proteins, which confirmed that A-RFP could enhance the docking performance. And the flexible-receptor virtual screening on a low-similarity dataset with 85 receptors validates the accuracy of residue flexibility comprehensive evaluation. Moreover, we studied three receptors with FDA-approved drugs, which further proved A-RFP can play a suitable role in ligand discovery. conclusion: Our analysis confirms that the screening performance of the various number of flexible residues varies wildly across receptors. It suggests that a fine-grained docking method would offset the aforementioned deficiency. Thus, we presented A-RFP, an adaptive pocket residue flexibility prediction method based on homologous information. Without considering computational resources and time costs, A-RFP provides the optimal docking result.

Prediction of large conformational changes of a protein binding pocket associated with ligand binding

Protein flexibility upon ligand binding: Docking predictions and statistical analysis

ApoDock: Ligand-Conditioned Sidechain Packing for Flexible Molecular Docking

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

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

DynamicBind: predicting ligand-specific protein-ligand complex structure with a deep equivariant generative model

Synergistic Application of Molecular Docking and Machine Learning for Improved Binding Pose

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

Pre-Training on Large-Scale Generated Docking Conformations with HelixDock to Unlock the Potential of Protein-ligand Structure Prediction Models

Predicting binding events in very flexible, allosteric, multi-domain proteins

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

Advances to tackle backbone flexibility in protein docking

Protein-ligand binding affinity prediction: Is 3D binding pose needed?

Author Correction: Integrated Structure Prediction of Protein–protein Docking with Experimental Restraints Using ColabDock

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

Prediction of ligand binding mode among multiple cross-docking poses by molecular dynamics simulations

E3Bind: An End-to-End Equivariant Network for Protein-Ligand Docking

ColabDock: Inverting AlphaFold Structure Prediction Model for Protein-Protein Docking with Experimental Restraints

A-RFP: An Adaptive Residue Flexibility Prediction Method Improving Protein-ligand Docking Based on Homologous Proteins

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

DeepBSP—a Machine Learning Method for Accurate Prediction of Protein–Ligand Docking Structures