A New Hybrid Neural Network Deep Learning Method for Protein–Ligand Binding Affinity Prediction and De Novo Drug Design

Abstract

Accurately predicting ligand binding affinity in a virtual screening campaign is still challenging. Here, we developed hybrid neural network (HNN) machine deep learning methods, HNN-denovo and HNN-affinity, by combining the 3D-CNN (convolutional neural network) and the FFNN (fast forward neural network) hybrid neural network framework. The HNN-denovo uses protein pocket structure and protein–ligand interactions as input features. The HNN-affinity uses protein sequences and ligand features as input features. The HNN method combines the CNN and FCNN machine architecture for the protein structure or protein sequence and ligand descriptors. To train the model, the HNN methods used thousands of known protein–ligand binding affinity data retrieved from the PDBBind database. We also developed the Random Forest (RF), Gradient Boosting (GB), Decision Tree with AdaBoost (DT), and a consensus model. We compared the HNN results with models developed based on the RF, GB, and DT methods. We also independently compared the HNN method results with the literature reported deep learning protein–ligand binding affinity predictions made by the DLSCORE, KDEEP, and DeepAtom. The predictive performance of the HNN methods (max Pearson’s R achieved was 0.86) was consistently better than or comparable to the DLSCORE, KDEEP, and DeepAtom deep learning learning methods for both balanced and unbalanced data sets. The HNN-affinity can be applied for the protein–ligand affinity prediction even in the absence of protein structure information, as it considers the protein sequence as standalone feature in addition to the ligand descriptors. The HNN-denovo method can be efficiently implemented to the structure-based de novo drug design campaign. The HNN-affinity method can be used in conjunction with the deep learning molecular docking protocols as a standalone. Further, it can be combined with the conventional molecular docking methods as a multistep approach to rapidly screen billions of diverse compounds. The HNN method are highly scalable in the cloud ML platform.

Prediction of protein–ligand binding affinity via deep learning models

“…Similarly, Limbu et al. developed a hybrid neural network–affinity model based on only ligand SMILES strings and protein sequences as input features [ 66 ]. In addition to the ligand SMILES strings and protein sequences, the interaction-free models, such as DeepDTAF [ 64 ], DEELIG [ 69 ], PLA-MoRe [ 65 ], PLANET [ 67 ] and CAPLA [ 81 ], also take the pocket sequence, the predicted secondary structure, atomic and residual physicochemical properties of protein/pocket, as well as the bioactive properties of the ligand as inputs.…”

Predicting Dose-Dependent Carcinogenicity of Chemical Mixtures Using a Novel Hybrid Neural Network Framework and Mathematical Approach

“…A vector was created for the SMILES of each compound by converting each character in the SMILES string to its corresponding index in the dictionary. The resulting vector for the SMILES was padded with zeros or truncated so that it was of uniform length, L. We previously described in detail the SMILES preprocessing of chemicals [ 76 , 77 , 78 ].…”

“…We utilized the hybrid neural network (HNN) framework [ 76 , 77 , 78 ], which we developed for single chemical toxicity and carcinogen prediction. In contrast to the original model’s one-hot encoding of SMILES, here, we vectorized the SMILES using the method described in the SMILES preprocessing section.…”

A Machine Learning-Driven Pathophysiology Based New Approach Method for the Dose-Dependent Assessment of Hazardous Chemical Mixtures and Experimental Validations.

“…In the last step, these descriptors were given as input features for the FFNN and CNN of the HNN hybrid framework, and simulations were initiated. The output of the CNN, and FFNN are merged, within the HNN framework [45][46][47][48] to create mixture classification models which are described below. Eventually, we predicted the unknown chemical mixture toxicity in a dosedependent manner with the inclusion of chemical interaction effect.…”

“…Consequently, the qualitative and quantitative data assessing the mixtures and subsequent adverse effects are lacking, making the translation of existing data into meaningful prevention and therapeutic strategies [7,. We recently described a AI hybrid neural network (AI-HNN) machine learning method for predicting the binary, multiclass and categorical carcinogenicity of chemicals and their mixtures in a dose-dependent manner [45][46][47][48]. However, this method does not account for post-exposure effects such as toxicokinetic and toxicodynamic properties, among other toxicological and pathophysiological characteristics, that limits the utility of this method.…”