Kejiang Lin

Precise prediction of bioactive conformations represents a central challenge in drug discovery. Traditional computational methods often rely on the assumption that the global minimum energy state corresponds to the bioactive conformation─a hypothesis with inherent limitations. Data-driven artificial intelligence (AI) models have revolutionized research methodologies in this domain. By learning the implicit probability distributions of conformations, AI demonstrates the potential to achieve high-precision predictions of bioactive structures. This review systematically surveys technical advancements ranging from geometric regression to generative and conditional generative models, analyzing their effectiveness in benchmarks and de novo design experiments. We deeply explore critical challenges, including data sparsity, sampling efficiency, generation capability, and interpretability. Finally, we provide an outlook on bridging the gap from "structural prediction" to "functional design" through the construction of dynamic data sets, the development of efficient algorithms, and the implementation of human-in-the-loop collaborative decision-making systems.

Background: The epidemic caused by SARS-CoV-2 swept the world in 2019. The 3C-like protease (3CLpro) of SARS-CoV-2 plays a key role in viral replication, and its inhibition could inhibit viral repl... Graphical abstract

Covalent drugs exhibit advantages in that noncovalent drugs cannot match, and covalent docking is an important method for screening covalent lead compounds. However, it is difficult for covalent docking to screen covalent compounds on a large scale because covalent docking requires determination of the covalent reaction type of the compound. Here, we propose to use deep learning of a lateral interactions spiking neural network to construct a covalent lead compound screening model to quickly screen covalent lead compounds. We used the 3CL protease (3CL Pro) of SARS-CoV-2 as the screen target and constructed two classification models based on LISNN to predict the covalent binding and inhibitory activity of compounds. The two classification models were trained on the covalent complex data set targeting cysteine (Cys) and the compound inhibitory activity data set targeting 3CL Pro, respected, with good prediction accuracy (ACC > 0.9). We then screened the screening compound library with 6 covalent binding screening models and 12 inhibitory activity screening models. We tested the inhibitory activity of the 32 compounds, and the best compound inhibited SARS-CoV-2 3CL Pro with an IC₅₀ value of 369.5 nM. Further assay implied that dithiothreitol can affect the inhibitory activity of the compound to 3CL Pro, indicating that the compound may covalently bind 3CL Pro. The selectivity test showed that the compound had good target selectivity to 3CL Pro over cathepsin L. These correlation assays can prove the rationality of the covalent lead compound screening model. Finally, covalent docking was performed to demonstrate the binding conformation of the compound with 3CL Pro. The source code can be obtained from the GitHub repository (https://github.com/guzh970630/Screen_Covalent_Compound_by_LISNN).

Identifying compound–protein interaction plays a vital role in drug discovery. Artificial intelligence (AI), especially machine learning (ML) and deep learning (DL) algorithms, are playing increasingly important roles in compound-protein interaction (CPI) prediction. However, ML relies on learning from large sample data. And the CPI for specific target often has a small amount of data available. To overcome the dilemma, we propose a virtual screening model, in which word2vec is used as an embedding tool to generate low-dimensional vectors of SMILES of compounds and amino acid sequences of proteins, and the modified multi-grained cascade forest based gcForest is used as the classifier. This proposed method is capable of constructing a model from raw data, adjusting model complexity according to the scale of datasets, especially for small scale datasets, and is robust with few hyper-parameters and without over-fitting. We found that the proposed model is superior to other CPI prediction models and performs well on the constructed challenging dataset. We finally predicted 2 new inhibitors for clusters of differentiation 47(CD47) which has few known inhibitors. The IC₅₀s of enzyme activities of these 2 new small molecular inhibitors targeting CD47-SIRPα interaction are 3.57 and 4.79 μM respectively. These results fully demonstrate the competence of this concise but efficient tool for CPI prediction.

Fibrosis is a disease that damages organs and even causes death. Because of the complicated pathogenesis, the development of drugs for fibrosis is challenging. In the lysophosphatidic acid receptor...

Cluster of differentiation 47 (CD47)/signal regulatory protein alpha (SIRPα) is a negative innate immune checkpoint signaling pathway that restrains immunosurveillance and immune clearance, and thus has aroused wide interest in cancer immunotherapy. Blockade of the CD47/SIRPα signaling pathway shows remarkable antitumor effects in clinical trials. Currently, all inhibitors targeting CD47/SIRPα in clinical trials are biomacromolecules. The poor permeability and undesirable oral bioavailability of biomacromolecules have caused researchers to develop small-molecule CD47/SIRPα pathway inhibitors. This review will summarize the recent advances in CD47/SIRPα interactions, including crystal structures, peptides and small molecule inhibitors. In particular, we have employed computer-aided drug discovery (CADD) approaches to analyze all the published crystal structures and docking results of small molecule inhibitors of CD47/SIRPα, providing insight into the key interaction information to facilitate future development of small molecule CD47/SIRPα inhibitors.

Virtual screening is an important means for lead compound discovery. The scoring function is the key to selecting hit compounds. Many scoring functions are currently available; however, there are no all-purpose scoring functions because different scoring functions tend to have conflicting results. Recently, neural networks, especially convolutional neural networks, have constantly been penetrating drug design and most CNN-based virtual screening methods are superior to traditional docking methods, such as Dock and AutoDock. CNNbased virtual screening is expected to improve the previous model of overreliance on computational chemical screening. Utilizing the powerful learning ability of neural networks provides us with a new method for evaluating compounds. We review the latest progress of CNN-based virtual screening and propose prospects.

With the rise of artificial intelligence (AI) in drug discovery, de novo molecular generation provides new ways to explore chemical space. However, because de novo molecular generation methods rely on abundant known molecules, generated molecules may have a problem of novelty. Novelty is important in highly competitive areas of medicinal chemistry, such as the discovery of kinase inhibitors. In this study, de novo molecular generation based on recurrent neural networks was applied to discover a new chemical space of kinase inhibitors. During the application, the practicality was evaluated, and new inspiration was found. With the successful discovery of one potent Pim1 inhibitor and two lead compounds that inhibit CDK4, AI-based molecular generation shows potentials in drug discovery and development.