Identification of novel protein binding sites expands druggable genome and opens new opportunities for drug discovery. Generally, presence or absence of a binding site depends on the three-dimensional conformation of a protein, making binding site identification resemble the object detection problem in computer vision. Here we introduce a computational approach for the large-scale detection of protein binding sites, that considers protein conformations as 3D-images, binding sites as objects on these images to detect, and conformational ensembles of proteins as 3D-videos to analyze. BiteNet is suitable for spatiotemporal detection of hard-to-spot allosteric binding sites, as we showed for conformation-specific binding site of the epidermal growth factor receptor, oligomer-specific binding site of the ion channel, and binding site in G protein-coupled receptor. BiteNet outperforms state-of-the-art methods both in terms of accuracy and speed, taking about 1.5 minutes to analyze 1000 conformations of a protein with ~2000 atoms.

Peptides and peptide-based molecules represent a promising therapeutic modality targeting intracellular protein−protein interactions, potentially combining the beneficial properties of biologics and small-molecule drugs. Protein−peptide complexes occupy a unique niche of interaction interfaces with respect to protein−protein and protein− small molecule complexes. Protein−peptide binding site identification resembles image object detection, a field that had been revolutionalized with computer vision techniques. We present a new protein−peptide binding site detection method called BiteNet Pp by harnessing the power of 3D convolutional neural network. Our method employs a tensor-based representation of spatial protein structures, which is fed to 3D convolutional neural network, resulting in probability scores and coordinates of the binding "hot spots" in the input structures. We used the domain adaptation technique to fine-tune model trained on protein−small molecule complexes using a manually curated set of protein−peptide structures. BiteNet Pp consistently outperforms existing state-of-the-art methods in the independent test benchmark. It takes less than a second to analyze a single-protein structure, making BiteNet Pp suitable for the large-scale analysis of protein− peptide binding sites.

Structure-based drug design (SBDD) targeting nucleic acid macromolecules, particularly RNA, is a gaining momentum research direction that already resulted in several FDA-approved compounds. Similar to proteins, one of the critical components in SBDD for RNA is the correct identification of the binding sites for putative drug candidates. RNAs share a common structural organization that, together with the dynamic nature of these molecules, makes it challenging to recognize binding sites for small molecules. Moreover, there is a need for structure-based approaches, as sequence information only does not consider conformation plasticity of nucleic acid macromolecules. Deep learning holds a great promise to resolve binding site detection problem, but requires a large amount of structural data, which is very limited for nucleic acids, compared to proteins. In this study we composed a set of ∼2000 nucleic acid-small molecule structures comprising ∼2500 binding sites, which is ∼40-times larger than previously used one, and demonstrated the first structure-based deep learning approach, BiteNetN, to detect binding sites in nucleic acid structures. BiteNetN operates with arbitrary nucleic acid complexes, shows the state-of-the-art performance, and can be helpful in the analysis of different conformations and mutant variants, as we demonstrated for HIV-1 TAR RNA and ATP-aptamer case studies.

No abstract

Identification of novel protein binding sites expands «druggable genome» and opens new opportunities for drug discovery. Generally, presence or absence of a binding site depends on the three-dimensional conformation of a protein, making binding site identification resemble to object detection problem in computer vision. Here we introduce a computational approach for the largescale detection of protein binding sites, named BiteNet, that considers protein conformations as the 3D-images, binding sites as the objects on these images to detect, and conformational ensembles of proteins as the 3D-videos to analyze. BiteNet is suitable for spatiotemporal detection of hard-tospot allosteric binding sites, as we showed for conformation-specific binding site of the epidermal growth factor receptor, oligomer-specific binding site of the ion channel, and binding sites in G protein-coupled receptors. BiteNet outperforms state-of-the-art methods both in terms of accuracy and speed, taking about 1.5 minute to analyze 1000 conformations of a protein with ∼ 2000 atoms. BiteNet is available at https://github.com/i-Molecule/bitenet.