Atomic Convolutional Networks for Predicting Protein-Ligand Binding Affinity

By Joseph Gomes, Bharath Ramsundar, Evan N. Feinberg, Vijay S. Pande. 2017

208 citations recorded by the source corpus.

Empirical scoring functions based on either molecular force fields or cheminformatics descriptors are widely used, in conjunction with molecular docking, during the early stages of drug discovery to predict potency and binding affinity of a drug-like molecule to a given target. These models require expert-level knowledge of physical chemistry and biology to be encoded as hand-tuned parameters or features rather than allowing the underlying model to select features in a data-driven procedure. Here, we develop a general 3-dimensional spatial convolution operation for learning atomic-level chemical interactions directly from atomic coordinates and demonstrate its application to structure-based bioactivity prediction. The atomic convolutional neural network is trained to predict the experimentally determined binding affinity of a protein-ligand complex by direct calculation of the energy associated with the complex, protein, and ligand given the crystal structure of the binding pose. Non-covalent interactions present in the complex that are absent in the protein-ligand sub-structures are identified and the model learns the interaction strength associated with these features. We test our model by predicting the binding free energy of a subset of protein-ligand complexes found in the PDBBind dataset and compare with state-of-the-art cheminformatics and machine learning-based approaches. We find that all methods achieve experimental accuracy and that atomic convolutional networks either outperform or perform competitively with the cheminformatics based methods. Unlike all previous protein-ligand prediction systems, atomic convolutional networks are end-to-end and fully-differentiable. They represent a new data-driven, physics-based deep learning model paradigm that offers a strong foundation for future improvements in structure-based bioactivity prediction.

Subfields: Computational Theory and Mathematics, Materials Chemistry, Molecular Biology, Graph Learning, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Materials Chemistry

In Molecular Biology

In Graph Learning

In Computational Chemistry & Drug Discovery

Citations in the corpus

Learning Graph-Level Representation for Drug Discovery (2017, background): “…et al. 2015), similar graph networks that take the original molecule as input(Altae-Tran et al. 2017; Yao et al. 2017; Schütt et al. 2017; Gomes et al. 2017), and other general graph neural networks designed for all types of graph(Li et al. 2015; Bruna et al. 2013; Defferrard, Bresson,…”
Opportunities and obstacles for deep learning in biology and medicine (2017, methodology): “Xu et al. [179] applied CNNs to the detection of enhancers, achieving incremental improvements in specificity and sensitivity over a previous SVM-based approach, and much better performance for cell-specific enhancers.”
DeepDTA: deep drug–target binding affinity prediction (2018, unknown): “Deep learning approaches have also been applied to protein-ligand interaction scoring in which a common application has been the use of CNNs that learn from the 3D structures of the protein-ligand complexes [21, 43, 51].”
End-to-End Learning on 3D Protein Structure for Interface Prediction (2018, methodology): “[10] uses the symmetry functions for protein-ligand binding affinity prediction.”
A high quality, industrial data set for binding affinity prediction: performance comparison in different early drug discovery scenarios (2022, methodology): “We further include two 3D-based ML models referred to as atomic convolutional neural network (ACNN) and PotentialNet [7, 18].”
Machine learning in protein engineering (2018, unknown): “…et al., 2015; Zeng et al., 2016), protein-MHC binding (Hu & Liu, 2017), binding site prediction (Jim ´ enez et al., 2017), protein-ligand binding (Gomes et al., 2017; Mazzaferro, 2017), solubility (Khurana et al., 2018), thermostability (Dehouck et al., 2009; Giollo et al., 2014), subcellular…”
Graph convolutional networks for computational drug development and discovery (2019, background): “Second, traditional methods often view interaction predication as a binary classification problem and are unable to handle cold-target problems [63, 64], i.”
Machine-learning-guided directed evolution for protein engineering (2018, background): “…to protein–nucleic acid binding 46–48 , protein–major histocompatibility complex binding 49 , binding-site prediction 50 , protein–ligand binding 51,52 , solubility 53 , thermostability 54,55 , subcellular localization 56 , secondary structure 57 , functional class 58–60 , and even…”
A Transdisciplinary Review of Deep Learning Research and Its Relevance for Water Resources Scientists (2017, methodology): “…(more) raw-level data seemed to perform on par with or better than those working with predefined features or descriptors (Goh, Siegel, et al., 2017; Gomes et al., 2017; Lusci et al., 2013; Mayr et al., 2016; S. L. Yang et al., 2014); (2) multitask learning may produce stronger models as they…”