Location and properties of ligand binding sites provide important information to uncover protein functions and to direct structure-based drug design approaches. However, as binding site detection depends on the three-dimensional (3D) structural data of proteins, functional analysis based on protein ligand binding sites is formidable for proteins without structural information. Recent developments in protein structure prediction and the 3D structures built by AlphaFold provide an unprecedented opportunity for analyzing ligand binding sites in human proteins. Here, we constructed the CavitySpace database, the first pocket library for all the proteins in the human proteome, using a widely-applied ligand binding site detection program CAVITY. Our analysis showed that known ligand binding sites could be well recovered. We grouped the predicted binding sites according to their similarity which can be used in protein function prediction and drug repurposing studies. Novel binding sites in highly reliable predicted structure regions provide new opportunities for drug discovery. Our CavitySpace is freely available and provides a valuable tool for drug discovery and protein function studies.
Deep generative models have advanced drug discovery but often generate compounds with limited structural novelty, providing constrained inspiration for medicinal chemists. To address this, we develop TransPharmer, a generative model that integrates ligand-based interpretable pharmacophore fingerprints with a generative pre-training transformer (GPT)-based framework for de novo molecule generation. TransPharmer excels in unconditioned distribution learning, de novo generation, and scaffold elaboration under pharmacophoric constraints. Its unique exploration mode could enhance scaffold hopping, producing structurally distinct but pharmaceutically related compounds. Its efficacy is validated through two case studies involving the dopamine receptor D2 (DRD2) and polo-like kinase 1 (PLK1). Notably, three out of four synthesized PLK1-targeting compounds show submicromolar activities, with the most potent, IIP0943, exhibiting a potency of 5.1 nM. Featuring a new 4-(benzo[b]thiophen-7-yloxy)pyrimidine scaffold, IIP0943 also has high PLK1 selectivity and submicromolar inhibitory activity in HCT116 cell proliferation. TransPharmer offers a promising tool for discovering structurally novel and bioactive ligands.
15 September 2022 in Journal of Chemical Information and Modeling
Molecular structures are commonly depicted in 2D printed forms in scientific documents such as journal papers and patents. However, these 2D depictions are not machine readable. Due to a backlog of decades and an increasing amount of printed literatures, there is a high demand for translating printed depictions into machine-readable formats, which is known as Optical Chemical Structure Recognition (OCSR). Most OCSR systems developed over the last three decades use a rule-based approach, which vectorizes the depiction based on the interpretation of vectors and nodes as bonds and atoms. Here, we present a practical software called MolMiner, which is primarily built using deep neural networks originally developed for semantic segmentation and object detection to recognize atom and bond elements from documents. These recognized elements can be easily connected as a molecular graph with a distance-based construction algorithm. MolMiner gave state-of-the-art performance on four benchmark data sets and a self-collected external data set from scientific papers. As MolMiner performed similarly well in real-world OCSR tasks with a user-friendly interface, it is a useful and valuable tool for daily applications. The free download links of Mac and Windows versions are available at https://github.com/iipharma/pharmamind-molminer.
The drug development pipeline involves several stages including in vitro assays, in vivo assays, and clinical trials. For candidate selection, it is important to consider that a compound will successfully pass through these stages. Using graph neural networks, we developed three subdivisional models to individually predict the capacity of a compound to enter in vivo testing, clinical trials, and market approval stages. Furthermore, we proposed a strategy combing both active learning and ensemble learning to improve the quality of the models. The models achieved satisfactory performance in the internal test datasets and four self-collected external test datasets. We also employed the models as a general index to make an evaluation on a widely known benchmark dataset DEKOIS 2.0, and surprisingly found a powerful ability on virtual screening tasks. Our model system (termed as miDruglikeness) provides a comprehensive drug-likeness prediction tool for drug discovery and development.
We developed MaxQsaring, a novel universal framework integrating molecular descriptors, fingerprints, and deep-learning pretrained representations, to predict the properties of compounds. Applied to a case study of human ether-à-go-go-related gene (hERG) blockage prediction, MaxQsaring achieved state-of-the-art performance on two challenging external datasets through automatic optimal feature combinations, and successfully identified top 10 important interpretable features that could be used to model a high-accuracy decision tree. The models' predictions align well with empirical hERG optimization strategies, demonstrating their interpretability for practical utilities. Deep learning pre-trained representations have been demonstrated to exert a moderate influence on enhancing the performance of predictive models. Nevertheless, their impact on augmenting the generalizability of these models, particularly when applied to compounds possessing novel scaffolds, appears to be comparatively minimal. MaxQsaring excelled in the Therapeutics Data Commons (TDC) benchmarks, ranking first in 19 out of 22 tasks, showcasing its potential for universal accurate compound property prediction to facilitate a high success rate of early drug discovery, which is still a formidable challenge.