Predicting protein-ligand binding residues with deep convolutional neural networks

Abstract

Background Ligand-binding proteins play key roles in many biological processes. Identification of protein-ligand binding residues is important in understanding the biological functions of proteins. Existing computational methods can be roughly categorized as sequence-based or 3D-structure-based methods. All these methods are based on traditional machine learning. In a series of binding residue prediction tasks, 3D-structure-based methods are widely superior to sequence-based methods. However, due to the great number of proteins with known amino acid sequences, sequence-based methods have considerable room for improvement with the development of deep learning. Therefore, prediction of protein-ligand binding residues with deep learning requires study. Results In this study, we propose a new sequence-based approach called DeepCSeqSite for ab initio protein-ligand binding residue prediction. DeepCSeqSite includes a standard edition and an enhanced edition. The classifier of DeepCSeqSite is based on a deep convolutional neural network. Several convolutional layers are stacked on top of each other to extract hierarchical features. The size of the effective context scope is expanded as the number of convolutional layers increases. The long-distance dependencies between residues can be captured by the large effective context scope, and stacking several layers enables the maximum length of dependencies to be precisely controlled. The extracted features are ultimately combined through one-by-one convolution kernels and softmax to predict whether the residues are binding residues. The state-of-the-art ligand-binding method COACH and some of its submethods are selected as baselines. The methods are tested on a set of 151 nonredundant proteins and three extended test sets. Experiments show that the improvement of the Matthews correlation coefficient (MCC) is no less than 0.05. In addition, a training data augmentation method that slightly improves the performance is discussed in this study. Conclusions Without using any templates that include 3D-structure data, DeepCSeqSite significantlyoutperforms existing sequence-based and 3D-structure-based methods, including COACH. Augmentation of the training sets slightly improves the performance. The model, code and datasets are available at https://github.com/yfCuiFaith/DeepCSeqSite .

Recognizing Ion Ligand–Binding Residues by Random Forest Algorithm Based on Optimized Dihedral Angle

“…According to the description in the literatures (Sodhi et al, 2004;Hu et al, 2016a,b;Cao et al, 2017;Li et al, 2017;Cui et al, 2019) and the statistical analysis of the position conservation in metal ion ligand-binding segments and nonbinding segments done by our group (Cao et al, 2017), we selected the amino acid position conservation information as a feature parameter. Because the dimension of PSSM (20 * L) that is commonly used to extract position conservation information is excessively high, we constructed position weight matrices to extract position conservation information of amino acids (Kel et al, 2003;Gao and Hu, 2014).…”

“…Therefore, some researchers selected dihedral angle as feature and obtained improved results. But the extraction method, using two-dimensional real values of phi and psi angles as features (Hu et al, 2016b;Cui et al, 2019), ignored character of dihedral angle of each ion ligandbinding residue. In this work, the phi and psi angles were performed by using statistical analysis and reclassification, and they were extracted as feature parameters.…”

Pathway-Specific Protein Domains (PSPD) Discrimination by Using a Hybrid Feature Space Based on Deep Neural Network(DNN)

“…Although DNNPSPDs can improve pathway-specific protein prediction accuracy and precision to some extent, its predictability and algorithm efficiency require further improvements. We shall also attempt to build our fundamental knowledge on proteins to produce more successful hidden features, to take biological meaning into account, and to incorporate specific effective algorithms, such as convolutional neural networks [65], capsule networks [66], and generative opponent networks. The versatility of our work contributes to the advancements in protein function analysis and association predictions with human pathways based on convolutional neural networks and several specific biological data sources, such as the graph embedding features or pathway involvement of protein-protein networks.…”

Predicting Protein-Ligand Binding Site for Drug Design Using Context Relevant Self Organizing Maps (CRSOM)

“…In (10), (11), and (12), 2 is the learning rate for reference vectors. is the backpropagated error information from the output layer.…”

“…Research using structure and sequence-based computing approaches to predict binding sites has been carried out, including: Predicting Functionally Important Residues from Sequence Conservation [11], Identification of Protein-Ligand Binding Sites by Sequence Information and Ensemble Classifier [12], and Integrating Data Selection and Extreme Learning Machine to Predict Protein-Ligand Binding site [2].…”