Large-Scale Validation of a Quantum Mechanics Based Scoring Function:  Predicting the Binding Affinity and the Binding Mode of a Diverse Set of Protein−Ligand Complexes

2005 ◽  
Vol 48 (14) ◽  
pp. 4558-4575 ◽  
Author(s):  
Kaushik Raha ◽  
Kenneth M. Merz
Keyword(s):  
Quantum Mechanics ◽  
Binding Affinity ◽  
Large Scale ◽  
Scale Validation ◽  
Scoring Function ◽  
Binding Mode

CSCORE: A SIMPLE YET EFFECTIVE SCORING FUNCTION FOR PROTEIN–LIGAND BINDING AFFINITY PREDICTION USING MODIFIED CMAC LEARNING ARCHITECTURE

2011 ◽  
Vol 09 (supp01) ◽  
pp. 1-14 ◽  
Author(s):  
XUCHANG OUYANG ◽  
STEPHANUS DANIEL HANDOKO ◽  
CHEE KEONG KWOH
Keyword(s):  
Binding Affinity ◽  
Scoring Function ◽  
Binding Mode ◽  
Computational Method ◽  
Data Driven ◽  
Machine Learning Techniques ◽  
Ligand Docking ◽  
Scoring Functions ◽  
Binding Affinity Prediction ◽  
Affinity Prediction

Protein–ligand docking is a computational method to identify the binding mode of a ligand and a target protein, and predict the corresponding binding affinity using a scoring function. This method has great value in drug design. After decades of development, scoring functions nowadays typically can identify the true binding mode, but the prediction of binding affinity still remains a major problem. Here we present CScore, a data-driven scoring function using a modified Cerebellar Model Articulation Controller (CMAC) learning architecture, for accurate binding affinity prediction. The performance of CScore in terms of correlation between predicted and experimental binding affinities is benchmarked under different validation approaches. CScore achieves a prediction with R = 0.7668 and RMSE = 1.4540 when tested on an independent dataset. To the best of our knowledge, this result outperforms other scoring functions tested on the same dataset. The performance of CScore varies on different clusters under the leave-cluster-out validation approach, but still achieves competitive result. Lastly, the target-specified CScore achieves an even better result with R = 0.8237 and RMSE = 1.0872, trained on a much smaller but more relevant dataset for each target. The large dataset of protein–ligand complexes structural information and advances of machine learning techniques enable the data-driven approach in binding affinity prediction. CScore is capable of accurate binding affinity prediction. It is also shown that CScore will perform better if sufficient and relevant data is presented. As there is growth of publicly available structural data, further improvement of this scoring scheme can be expected.

scholarly journals DeepBindRG: a deep learning based method for estimating effective protein–ligand affinity

PeerJ ◽  
2019 ◽  
Vol 7 ◽  
pp. e7362 ◽  
Author(s):  
Haiping Zhang ◽  
Linbu Liao ◽  
Konda Mani Saravanan ◽  
Peng Yin ◽  
Yanjie Wei
Keyword(s):  
Deep Learning ◽  
Ligand Binding ◽  
Binding Affinity ◽  
Mean Squared Error ◽  
Scoring Function ◽  
Binding Mode ◽  
Scoring Functions ◽  
Autodock Vina ◽  
Cellular Functions ◽  
Ligand Complex

Proteins interact with small molecules to modulate several important cellular functions. Many acute diseases were cured by small molecule binding in the active site of protein either by inhibition or activation. Currently, there are several docking programs to estimate the binding position and the binding orientation of protein–ligand complex. Many scoring functions were developed to estimate the binding strength and predict the effective protein–ligand binding. While the accuracy of current scoring function is limited by several aspects, the solvent effect, entropy effect, and multibody effect are largely ignored in traditional machine learning methods. In this paper, we proposed a new deep neural network-based model named DeepBindRG to predict the binding affinity of protein–ligand complex, which learns all the effects, binding mode, and specificity implicitly by learning protein–ligand interface contact information from a large protein–ligand dataset. During the initial data processing step, the critical interface information was preserved to make sure the input is suitable for the proposed deep learning model. While validating our model on three independent datasets, DeepBindRG achieves root mean squared error (RMSE) value of pKa (−logKd or −logKi) about 1.6–1.8 and R value around 0.5–0.6, which is better than the autodock vina whose RMSE value is about 2.2–2.4 and R value is 0.42–0.57. We also explored the detailed reasons for the performance of DeepBindRG, especially for several failed cases by vina. Furthermore, DeepBindRG performed better for four challenging datasets from DUD.E database with no experimental protein–ligand complexes. The better performance of DeepBindRG than autodock vina in predicting protein–ligand binding affinity indicates that deep learning approach can greatly help with the drug discovery process. We also compare the performance of DeepBindRG with a 4D based deep learning method “pafnucy”, the advantage and limitation of both methods have provided clues for improving the deep learning based protein–ligand prediction model in the future.

scholarly journals Nucleic binding affinity of bacteriophage T4 gene 32 protein in the cooperative binding mode.

1982 ◽  
Vol 257 (11) ◽  
pp. 6184-6193
Author(s):  
A M Bobst ◽  
P W Langemeier ◽  
P E Warwick-Koochaki ◽  
E V Bobst ◽  
J C Ireland
Keyword(s):  
Binding Affinity ◽  
Bacteriophage T4 ◽  
Binding Mode ◽  
Cooperative Binding ◽  
Gene 32 ◽  
T4 Gene 32 Protein

scholarly journals Structural basis of subunit selectivity for competitive NMDA receptor antagonists with preference for GluN2A over GluN2B subunits

2017 ◽  
Vol 114 (33) ◽  
pp. E6942-E6951 ◽  
Author(s):  
Genevieve E. Lind ◽  
Tung-Chung Mou ◽  
Lucia Tamborini ◽  
Martin G. Pomper ◽  
Carlo De Micheli ◽  
...  
Keyword(s):  
Nmda Receptor ◽  
Binding Affinity ◽  
Binding Mode ◽  
Structural Basis ◽  
Nmda Receptor Antagonists ◽  
Receptor Antagonists ◽  
Glutamate Binding ◽  
Subunit Interface ◽  
The Central Nervous System ◽  
Contact Residues

NMDA-type glutamate receptors are ligand-gated ion channels that contribute to excitatory neurotransmission in the central nervous system (CNS). Most NMDA receptors comprise two glycine-binding GluN1 and two glutamate-binding GluN2 subunits (GluN2A–D). We describe highly potent (S)-5-[(R)-2-amino-2-carboxyethyl]-4,5-dihydro-1H-pyrazole-3-carboxylic acid (ACEPC) competitive GluN2 antagonists, of which ST3 has a binding affinity of 52 nM at GluN1/2A and 782 nM at GluN1/2B receptors. This 15-fold preference of ST3 for GluN1/2A over GluN1/2B is improved compared with NVP-AAM077, a widely used GluN2A-selective antagonist, which we show has 11-fold preference for GluN1/2A over GluN1/2B. Crystal structures of the GluN1/2A agonist binding domain (ABD) heterodimer with bound ACEPC antagonists reveal a binding mode in which the ligands occupy a cavity that extends toward the subunit interface between GluN1 and GluN2A ABDs. Mutational analyses show that the GluN2A preference of ST3 is primarily mediated by four nonconserved residues that are not directly contacting the ligand, but positioned within 12 Å of the glutamate binding site. Two of these residues influence the cavity occupied by ST3 in a manner that results in favorable binding to GluN2A, but occludes binding to GluN2B. Thus, we reveal opportunities for the design of subunit-selective competitive NMDA receptor antagonists by identifying a cavity for ligand binding in which variations exist between GluN2A and GluN2B subunits. This structural insight suggests that subunit selectivity of glutamate-site antagonists can be mediated by mechanisms in addition to direct contributions of contact residues to binding affinity.

Water Mapping and Scoring approaches to predict the role of Hydration sites in Binding Affinity of PAK1 inhibitors

2021 ◽  
Vol 24 ◽  
Author(s):  
Jayashree Biswal ◽  
Prajisha Jayaprakash ◽  
Suresh Kumar Rayala ◽  
Ganesh Venkatraman ◽  
Raghu Rangasamy ◽  
...  
Keyword(s):  
Molecular Dynamics ◽  
Computational Model ◽  
Active Compound ◽  
Scoring Function ◽  
Binding Mode ◽  
Interaction Pattern ◽  
Dynamics Simulation ◽  
Interaction Patterns ◽  
Highly Active ◽  
Group I

Aim: This study aims to develop and establish a computational model that can identify potent molecules for p21-activating kinase 1 (PAK1). Background: PAK1 is a well-established drug target that has been explored for various therapeutic interventions. Control of this protein requires an indispensable inhibitor to curb the structural changes and subsequent activation of signalling effectors responsible for the progression of diseases, such as cancer, inflammatory, viral, and neurological disorders. Objective: To establish a computational model that could identify active molecules which will further provide a platform for developing potential PAK1 inhibitors. Method: A congeneric series of 27 compounds was considered for this study with Ki (nm) covering a minimum of 3 log range. The compounds were developed based on a previously reported Group-I PAK inhibitor, namely G-5555. The 27 compounds were subjected to the SP and XP mode of docking, to understand the binding mode, its conformation and interaction patterns. To understand the relevance of biological activity from computational approaches, the compounds were scored against generated water maps to obtain WM/MM ΔG binding energy. Moreover, molecular dynamics analysis was performed for the highly active compound, to understand the conformational variability and complex’s stability. We then evaluate the predictable binding pose obtained from the docking studies. Result: From the SP and XP modes of docking, the common interaction pattern with the amino acid residues Arg299 (cation-π), Glu345 (Aromatic hydrogen bond), hinge region Leu347, salt bridges Asp393 and Asp407 was observed, among the congeneric compounds. The interaction pattern was compared with the co-crystal inhibitor FRAX597 of the PAK1 crystal structure (PDB id: 4EQC). The correlation with different docking parameters in the SP and XP modes was insignificant and thereby revealed that the SP and XP’s scoring functions could not predict the active compounds. This was due to the limitations in the docking methodology that neglected the receptor flexibility and desolvation parameters. Hence, to recognise the desolvation and explicit solvent effects, as well as to study the Structure-Activity Relationships (SARs) extensively, WaterMap (WM) calculations were performed on the congeneric compounds. Based on displaceable unfavourable hydration sites (HS) and their associated thermodynamic properties, the WM calculations facilitated to understand the significance of correlation in the folds of activity of highly (19 and 17), moderate (16 and 21) and less active (26 and 25) compounds. Furthermore, the scoring function from WaterMap, namely WM/MM, led to a significant R2 value of 0.72, due to a coupled conjunction with MM treatment and displaced unfavourable waters at the binding site. To check the “optimal binding conformation”, molecular dynamics simulation was carried out with the highly active compound 19 to explain the binding mode, stability, interactions, solvent accessible area, etc., which could support the predicted conformation with bioactive conformation. Conclusion: This study determined the best scoring function, established SARs and predicted active molecules through a computational model. This will contribute towards development of the most potent PAK1 inhibitors.

Sign in / Sign up

Export Citation Format

Share Document