Current strategies to address data scarcity in artificial intelligence-based drug discovery: A comprehensive review.

Artificial intelligence (AI) has played a vital role in computer-aided drug design (CADD). This development has been further accelerated with the increasing use of machine learning (ML), mainly deep learning (DL), and computing hardware and software advancements. As a result, initial doubts about the application of AI in drug discovery have been dispelled, leading to significant benefits in medicinal chemistry. At the same time, it is crucial to recognize that AI is still in its infancy and faces a few limitations that need to be addressed to harness its full potential in drug discovery. Some notable limitations are insufficient, unlabeled, and non-uniform data, the resemblance of some AI-generated molecules with existing molecules, unavailability of inadequate benchmarks, intellectual property rights (IPRs) related hurdles in data sharing, poor understanding of biology, focus on proxy data and ligands, lack of holistic methods to represent input (molecular structures) to prevent pre-processing of input molecules (feature engineering), etc. The major component in AI infrastructure is input data, as most of the successes of AI-driven efforts to improve drug discovery depend on the quality and quantity of data, used to train and test AI algorithms, besides a few other factors. Additionally, data-gulping DL approaches, without sufficient data, may collapse to live up to their promise. Current literature suggests a few methods, to certain extent, effectively handle low data for better output from the AI models in the context of drug discovery. These are transferring learning (TL), active learning (AL), single or one-shot learning (OSL), multi-task learning (MTL), data augmentation (DA), data synthesis (DS), etc. One different method, which enables sharing of proprietary data on a common platform (without compromising data privacy) to train ML model, is federated learning (FL). In this review, we compare and discuss these methods, their recent applications, and limitations while modeling small molecule data to get the improved output of AI methods in drug discovery. Article also sums up some other novel methods to handle inadequate data.

Performance determinants and evidence-based practice in track cycling: A survey of coaches, practitioners, and athletes.

OBJECTIVES: This study examined how track cycling coaches, practitioners, and athletes: develop knowledge and practices; value performance areas; and, implement research into practice. DESIGN: Cross-sectional survey. METHODS: An online REDCap survey of track cycling coaches, practitioners, and athletes was conducted involving questions related to demographics, performance area importance, knowledge acquisition and application, research relevance, and research direction. RESULTS: A total of 159 responses were received from coaches (n = 55), practitioners (n = 29), and athletes (n = 75). Participants' highest track cycling competition level involvement ranged from local/regional (12.7%) to Olympic/Paralympic (39.9%). Respondents primarily develop practices by observing 'the sport' or 'others competing/working in it' (both 85.8%). Practitioners develop practices through self-guided learning (96.4%). The primary reason for practice use was prior experience (84.9%), whilst individuals were least likely to use practices resulting in marginal gains with potentially negative outcomes (27.3%). Areas of greatest perceived importance were Aerodynamics, Strength & Conditioning, and Tactics (all >96% agreed/strongly agreed). Scientific evidence for Tactics (30%) and Mental Skills (26%) was perceived to be lacking, resulting in greater reliance on personal experience (74% and 62%, respectively) to inform training decisions. The main barrier to implementing research into practice was athlete buy-in (84.3%). CONCLUSIONS: Within track cycling, informal learning was most popular amongst respondents. Greater reliance on personal experience within evidence-based practice for many performance areas aligns with limited existing research. Most respondents reported multiple barriers affecting research implementation in practice.

Many locks to one key: N-acetylneuraminic acid binding to proteins.

Sialic acids play crucial roles in cell surface glycans of both eukaryotic and prokaryotic organisms, mediating various biological processes, including cell-cell interactions, development, immune response, oncogenesis and host-pathogen interactions. This review focuses on the ß-anomeric form of N-acetylneuraminic acid (Neu5Ac), particularly its binding affinity towards various proteins, as elucidated by solved protein structures. Specifically, we delve into the binding mechanisms of Neu5Ac to proteins involved in sequestering and transporting Neu5Ac in Gram-negative bacteria, with implications for drug design targeting these proteins as antimicrobial agents. Unlike the initial assumptions, structural analyses revealed significant variability in the Neu5Ac binding pockets among proteins, indicating diverse evolutionary origins and binding modes. By comparing these findings with existing structures from other systems, we can effectively highlight the intricate relationship between protein structure and Neu5Ac recognition, emphasizing the need for tailored drug design strategies to inhibit Neu5Ac-binding proteins across bacterial species.

From Gut Microbiomes to Infectious Pathogens: Neurological Disease Game Changers.

Gut microbiota and infectious diseases affect neurological disorders, brain development, and function. Compounds generated in the gastrointestinal system by gut microbiota and infectious pathogens may mediate gut-brain interactions, which may circulate throughout the body and spread to numerous organs, including the brain. Studies shown that gut bacteria and disease-causing organisms may pass molecular signals to the brain, affecting neurological function, neurodevelopment, and neurodegenerative diseases. This article discusses microorganism-producing metabolites with neuromodulator activity, signaling routes from microbial flora to the brain, and the potential direct effects of gut bacteria and infectious pathogens on brain cells. The review also considered the neurological aspects of infectious diseases. The infectious diseases affecting neurological functions and the disease modifications have been discussed thoroughly. Recent discoveries and unique insights in this perspective need further validation. Research on the complex molecular interactions between gut bacteria, infectious pathogens, and the CNS provides valuable insights into the pathogenesis of neurodegenerative, behavioral, and psychiatric illnesses. This study may provide insights into advanced drug discovery processes for neurological disorders by considering the influence of microbial communities inside the human body.

Advancing cyanobacterial biotechnology: new frontiers in natural product discovery and production.

Cyanobacteria, as oxygenic phototrophs, offer significant potential for sustainable biotechnology applications. Cyanobacterial natural products, with antimicrobial, anticancer, and plant growth-promoting properties, hold promise in pharmaceuticals, agriculture, and environmental remediation. By leveraging advanced technologies, cyanobacteria can significantly impact various industries, supporting the green biotechnology agenda. Recent advancements in integrated omics, orphan gene cluster activation, genetic manipulation, and chemo-enzymatic methods are expanding their biotechnological relevance. Omics technologies revolutionize cyanobacterial natural product research by facilitating biosynthetic gene cluster identification. Heterologous expression and pathway reconstitution enable complex natural product production, while high-titer strategies like metabolic engineering enhance yields. Interdisciplinary research and technological progress position cyanobacteria as valuable sources of bioactive compounds, driving sustainable biotechnological practices forward.

Evaluating the harmonization potential of oral health-related questionnaires in national longitudinal birth and child cohort surveys.

BACKGROUND/OBJECTIVES: Effective use of longitudinal study data is challenging because of divergences in the construct definitions and measurement approaches over time, between studies and across disciplines. One approach to overcome these challenges is data harmonization. Data harmonization is a practice used to improve variable comparability and reduce heterogeneity across studies. This study describes the process used to evaluate the harmonization potential of oral health-related variables across each survey wave. METHODS: National child cohort surveys with similar themes/objectives conducted in the last two decades were selected. The Maelstrom Research Guidelines were followed for harmonization potential evaluation. RESULTS: Seven nationally representative child cohort surveys were included and questionnaires examined from 50 survey waves. Questionnaires were classified into three domains and fifteen constructs and summarized by age groups. A DataSchema (a list of core variables representing the suitable version of the oral health outcomes and risk factors) was compiled comprising 42 variables. For each study wave, the potential (or not) to generate each DataSchema variable was evaluated. Of the 2100 harmonization status assessments, 543 (26%) were complete. Approximately 50% of the DataSchema variables can be generated across at least four cohort surveys while only 10% (n = 4) variables can be generated across all surveys. For each survey, the DataSchema variables that can be generated ranged between 26% and 76%. CONCLUSION: Data harmonization can improve the comparability of variables both within and across surveys. For future cohort surveys, the authors advocate more consistency and standardization in survey questionnaires within and between surveys.

Combining Partial True Discovery Guarantee Procedures.

Closed testing has recently been shown to be optimal for simultaneous true discovery proportion control. It is, however, challenging to construct true discovery guarantee procedures in such a way that it focuses power on some feature sets chosen by users based on their specific interest or expertise. We propose a procedure that allows users to target power on prespecified feature sets, that is, "focus sets." Still, the method also allows inference for feature sets chosen post hoc, that is, "nonfocus sets," for which we deduce a true discovery lower confidence bound by interpolation. Our procedure is built from partial true discovery guarantee procedures combined with Holm's procedure and is a conservative shortcut to the closed testing procedure. A simulation study confirms that the statistical power of our method is relatively high for focus sets, at the cost of power for nonfocus sets, as desired. In addition, we investigate its power property for sets with specific structures, for example, trees and directed acyclic graphs. We also compare our method with AdaFilter in the context of replicability analysis. The application of our method is illustrated with a gene ontology analysis in gene expression data.

Identification of potential PIM-2 inhibitors via ligand-based generative models, molecular docking and molecular dynamics simulations.

Proviral Integrations of Moloney-2 (PIM-2) kinase is a promising target for various cancers and other diseases, and its inhibitors hold potential for treating related diseases. However, there is currently no clinically available PIM-2 inhibitor. In this study, we constructed a generative model for de novo PIM-2 inhibitor design based on artificial intelligence, performed molecular docking and molecular dynamics (MD) simulations to develop an efficient PIM-2 inhibitor generative model and discover potential PIM-2 inhibitors. First, we designed a generative model based on a Bi-directional Long Short-Term Memory (BiLSTM) framework combined with a transfer learning strategy and generated a new PIM-2 small molecule library using existing active drug databases. The generated compound library was then virtually screened by molecular docking and scaffold similarity comparison, identifying 10 initial hit compounds with better performance. Next, using the inhibitor in the crystal structure as a positive control, we performed two rounds of MD simulations, with lengths of 100 ns and 500 ns, respectively, to study the dynamic stability of the protein-ligand systems of the 10 compounds with PIM-2. Analyzed the interactions with key hinge region residues, binding free energies, and changes in the ATP pocket size. The generative model demonstrates good molecular generation capability and can generate efficient novel molecules with similar physicochemical properties as active PIM-2 drugs. Among the 10 initially selected hit compounds, 5 compounds C3 (- 29.69 kcal/mol), C4 (- 33.31 kcal/mol), C5 (- 28.59 kcal/mol), C8 (- 34.68 kcal/mol), and C9 (- 25.88 kcal/mol) have higher binding energies with PIM-2 than the positive drug 3YR (- 26.18 kcal/mol). The MD simulation results are consistent with the docking analysis, these compounds have lower and more stable RMSD values for the complex systems with the reported positive drug 3YR and PIM-2 complex system. They can form long-term stable interactions with active site and the hinge region of PIM-2, which suggests these compounds are likely to have potent inhibitory effects on PIM-2. This study provides an efficient generative model for PIM-2 inhibitor research and discovers 5 potential novel PIM-2 inhibitors.

Preclinical antidiabetic and antioxidant effects of Erythrophleum africanum (benth.) harms in streptozotocin-induced diabetic nephropathy.

OBJECTIVES: This study investigated the antidiabetic effects of the methanolic extract of E. africanum (MEEA) stem bark on streptozotocin (STZ)-induced diabetic nephropathy (DN) in Wistar rats. METHODS: The in vitro enzyme (α-amylase) inhibitory activity of MEEA was measured using a standard procedure. Diabetic rats with fasting blood glucose above 250 mg/dL were considered diabetic and were divided into the following groups: control (distilled water-treated), diabetic-control, diabetic metformin (100 mg/kg), diabetes + MEEA (150 mg/kg), and diabetes + MEEA (300 mg/kg) via oral gavage once daily for 14 days. At the end of the experimental period, kidney tissues were collected for biochemical and histological analyses. Kidney apoptosis and marker gene expression were measured by real-time quantitative PCR. RESULTS: MEEA exhibited α-amylase inhibitory effects. MEEA significantly (p<0.05) reduced the STZ-induced increases in blood glucose, serum urea, serum creatinine, uric acid, alanine aminotransferase, alkaline phosphatase, and malondialdehyde and increased the STZ-induced decreases in superoxide dismutase, catalase, and reduced glutathione. In addition, MEEA protects against DN by significantly downregulating the mRNA expression of cyclic adenosine monophosphate (cAMP), protein kinase A (PKA), cAMP-response binding protein (CREB), and cFOS and upregulating B-cell lymphoma 2 (Bcl-2), suggesting that the nephroprotective ability of MEEA is due to the modulation of the cAMP/PKA/CREB/cFOS signaling pathway. Furthermore, MEEA treatment protected against histopathological alterations observed in diabetic rats. CONCLUSIONS: The data from this study suggest that MEEA modulates glucose homeostasis and inhibits redox imbalance in DN rats.

SB-Net: Synergizing CNN and LSTM networks for uncovering retrosynthetic pathways in organic synthesis.

Retrosynthesis is vital in synthesizing target products, guiding reaction pathway design crucial for drug and material discovery. Current models often neglect multi-scale feature extraction, limiting efficacy in leveraging molecular descriptors. Our proposed SB-Net model, a deep-learning architecture tailored for retrosynthesis prediction, addresses this gap. SB-Net combines CNN and Bi-LSTM architectures, excelling in capturing multi-scale molecular features. It integrates parallel branches for processing one-hot encoded descriptors and ECFP, merging through dense layers. Experimental results demonstrate SB-Net's superiority, achieving 73.6 % top-1 and 94.6 % top-10 accuracy on USPTO-50k data. Versatility is validated on MetaNetX, with rates of 52.8 % top-1, 74.3 % top-3, 79.8 % top-5, and 83.5 % top-10. SB-Net's success in bioretrosynthesis prediction tasks indicates its efficacy. This research advances computational chemistry, offering a robust deep-learning model for retrosynthesis prediction. With implications for drug discovery and synthesis planning, SB-Net promises innovative and efficient pathways.

International Union of Basic and Clinical Pharmacology. CXV: The Class F of G Protein-Coupled Receptors.

The class F of G protein-coupled receptors (GPCRs) consists of ten Frizzleds (FZD1-10) and Smoothened (SMO). FZDs bind and are activated by secreted lipoglycoproteins of the Wingless/Int-1 (WNT) family and SMO is indirectly activated by the Hedgehog (Hh) family of morphogens acting on the transmembrane protein Patched (PTCH). The advance of our understanding of FZDs and SMO as dynamic transmembrane receptors and molecular machines, which emerged during the past 14 years since the first class F GPCR IUPHAR nomenclature report, justifies an update. This article focuses on the advances in molecular pharmacology and structural biology providing new mechanistic insight into ligand recognition, receptor activation mechanisms, signal initiation and signal specification. Furthermore, class F GPCRs continue to develop as drug targets, and novel technologies and tools such as genetically encoded biosensors and CRISP/Cas9 edited cell systems have contributed to refined functional analysis of these receptors. Also, advances in crystal structure analysis and cryogenic electron microscopy contribute to a rapid development of our knowledge about structure-function relationships providing a great starting point for drug development. Despite the progress questions and challenges remain to fully understand the complexity of the WNT/FZD and Hh/SMO signaling systems. Significance Statement The recent years of research have brought about substantial functional and structural insight into mechanisms of activation of Frizzleds and Smoothened. While the advance furthers our mechanistic understanding of ligand recognition, receptor activation, signal specification and initiation, broader opportunities emerge that allow targeting class F GPCRs for therapy and regenerative medicine employing both biologics and small molecule compounds.

Identifying ligands directly interacting with target protein in medicinal herbs by metabolomic analysis of T2-filtered HSQC spectra.

A protocol for efficiently identifying ligands directly interacting with a target protein in complex extracts of medicinal herbs was proposed by combining an adapted 2D perfect-echo Carr-Purcell-Meiboom-Gill heteronuclear single quantum correlation (PE-CPMG HSQC) spectrum with metabolomic analysis. PE-CPMG HSQC can suppress the signal interference from the target protein, allowing more accurate peak quantification than conventional HSQC. Inspired from untargeted metabolomics, regions of interest (ROIs) are constructed and quantified for the mixture or complex extract samples with and without a target protein, and then a binding index (BI) of each ROI is determined. ROIs or corresponding peaks significantly perturbed by the presence of the target protein (BI ≥1.5) are detected as differential features, and potential binding ligands identified from the differential features can be equated with bioactive markers associated with the 'treatment' of the target protein. Quantifying ROI can inclusively report the ligand bindings to a target protein in fast, intermediate and slow exchange regimes on nuclear magnetic resonance (NMR) time scale. The approach was successfully implemented and identified Angoroside C, Cinnamic acid and Harpagoside from the extract of Scrophularia ningpoensis Hemsl. as ligands binding to peroxisome proliferator-activated receptor γ. The proposed 2D NMR-based approach saves excess steps for sample processing and has a higher chance of detecting the weaker ligands in the complex extracts of medicinal herbs. We expect that this approach can be applied as an alternative to mining the potential ligands binding to a variety of target proteins from traditional Chinese medicines and herbal extracts.

Altered expression of cytokines, chemokines, growth factors, and soluble receptors in patients with colorectal cancer, and correlation with treatment outcome.

Insofar as they play an important role in the pathogenesis of colorectal cancer (CRC), this study analyzes the serum profile of cytokines, chemokines, growth factors, and soluble receptors in patients with CRC and cancer-free controls as possible CRC signatures. Serum levels of 65 analytes were measured in patients with CRC and age- and sex-matched cancer-free controls using the ProcartaPlex Human Immune Monitoring 65-Plex Panel. Of the 65 tested analytes, 8 cytokines (CSF-3, IFN-γ, IL-12p70, IL-18, IL-20, MIF, TNF-α and TSLP), 8 chemokines (fractalkine, MIP-1ß, BLC, Eotaxin-1, Eotaxin-2, IP-10, MIP-1a, MIP-3a), 2 growth factors (FGF-2, MMP-1), and 4 soluble receptors (APRIL, CD30, TNFRII, and TWEAK), were differentially expressed in CRC. ROC analysis confirmed the high association of TNF-α, BLC, Eotaxin-1, APRIL, and Tweak with AUC > 0.70, suggesting theranostic application. The expression of IFN-γ, IL-18, MIF, BLC, Eotaxin-1, Eotaxin-2, IP-10, and MMP1 was lower in metastatic compared to non-metastatic CRC; only AUC of MIF and MIP-1ß were > 0.7. Moreover, MDC, IL-7, MIF, IL-21, and TNF-α are positively associated with tolerance to CRC chemotherapy (CT) (AUC > 0.7), whereas IL-31, Fractalkine, Eotaxin-1, and Eotaxin-2 were positively associated with resistance to CT. TNF-α, BLC, Eotaxin-1, APRIL, and Tweak may be used as first-line early detection of CRC. The variable levels of MIF and MIP-1ß between metastatic and non-metastatic cases assign prognostic nature to these factors in CRC progression. Regarding tolerance to CT, MDC, IL-7, MIF, IL-21, and TNF-α are key when down-regulated or resistant to treatment is observed.

Gut Subdoligranulum variabile ameliorates rheumatoid arthritis by promoting TSG-6 synthesis from joint cells.

Background: A burgeoning body of evidence has substantiated the association between alterations in the composition of the gut microbiota and rheumatoid arthritis (RA). Nevertheless, our understanding of the intricate mechanisms underpinning this association is limited. Methods: To investigate whether the gut microbiota influences the pathogenesis of RA through metabolism or immunity, we performed rigorous synthesis analyses using aggregated statistics from published genome-wide association studies (GWAS) using two-sample Mendelian randomization (MR) and mediated MR techniques, including two-step MR and multivariate MR analyses. Subsequently, we conducted in vitro cellular validation of the analyzed Microbial-Cytokine-RA pathway. We determined the optimal culture conditions through co-culture experiments involving concentration and time. Cell Counting Kit-8 (CCK-8) assays were employed to assess cellular viability, and enzyme-linked immunosorbent assays (ELISA) were performed to assess tumor necrosis factor-inducible gene 6 protein (TSG-6) and tumor necrosis factor-α (TNF-α) levels. Results: Our univariable MR results confirmed 15 microbial traits, 7 metabolites and 2 cytokines that may be causally associated with RA (P FDR < 0.05). Mediation analysis revealed that microbial traits influence the risk of RA through metabolite or cytokine (proportion mediated: 7.75% - 58.22%). In vitro experiments demonstrated that TSG-6 was highly expressed in the Subdoligranulum variabile treatment group and was correlated with decreased RA severity (reduced TNF-α expression). Silencing the TSG-6 gene significantly increased TNF-α expression, regardless of treatment with S. variabile. Additionally, S. variabile-secreted exosomes exhibited the same effect. Conclusion: The results of this study suggest that S. variabile has the potential to promote TSG-6 secretion, thereby reducing RA inflammation.

Structure-based mapping of the histone-binding pocket of KDM4D using functionalized tetrazole and pyridine core compounds.

KDM4 histone demethylases became an exciting target for inhibitor development as the evidence linking them directly to tumorigenesis mounts. In this study, we set out to better understand the binding cavity using an X-ray crystallographic approach to provide a detailed landscape of possible interactions within the under-investigated region of KDM4. Our design strategy was based on utilizing known KDM binding motifs, such as nicotinic acid and tetrazolylhydrazides, as core motifs that we decided to enrich with flexible tails to map the distal histone binding site. The resulting X-ray structures of the novel compounds bound to KDM4D, a representative of the KDM4 family, revealed the interaction pattern with distal residues in the histone-binding site. The most prominent protein rearrangement detected upon ligand binding is the loop movement that blocks the accessibility to the histone binding site. Apart from providing new sites that potential inhibitors can target, the novel compounds may prove helpful in exploring the capacity of ligands to bind in sites distal to the cofactor-binding site of other KDMs or 2-oxoglutarate (2OG)-dependent oxygenases. The case study proves that combining a strong small binding motif with flexible tails to probe the binding pocket will facilitate lead discovery in classical drug-discovery campaigns, given the ease of accessing X-ray quality crystals.

Complete genome sequence of a novel amalgavirus in sponge gourd, Luffa cylindrica.

A novel monopartite dsRNA virus, tentatively named "sponge gourd amalgavirus 1" (SGAV1), was discovered by high-throughput sequencing in sponge gourd (Luffa cylindrica) displaying mosaic symptoms in Jiashan County, Zhejiang Province, China. The genome of SGAV1 is 3,447 nucleotides in length and contains partially overlapping open reading frames (ORFs) encoding a putative replication factory matrix-like protein and a fusion protein, respectively. The fusion protein of SGAV1 shares 57.07% identity with the homologous protein of salvia miltiorrhiza amalgavirus 1 (accession no. DAZ91057.1). Phylogenetic analysis based on the RNA-dependent RNA polymerase (RdRp) protein suggests that SGAV1 belongs to the genus Amalgavirus of the family Amalgaviridae. Moreover, analysis of SGAV1-derived small interfering RNAs indicated that SGAV1 was actively replicating in the host plant. Semi-quantitative RT-PCR showed higher levels of SGAV1 expression in leaves than in flowers and fruits. This is the first report of a novel amalgavirus found in sponge gourd in China.

Performance Assessment of Universal Machine Learning Interatomic Potentials: Challenges and Directions for Materials' Surfaces.

Machine learning interatomic potentials (MLIPs) are one of the main techniques in the materials science toolbox, able to bridge ab initio accuracy with the computational efficiency of classical force fields. This allows simulations ranging from atoms, molecules, and biosystems, to solid and bulk materials, surfaces, nanomaterials, and their interfaces and complex interactions. A recent class of advanced MLIPs, which use equivariant representations and deep graph neural networks, is known as universal models. These models are proposed as foundation models suitable for any system, covering most elements from the periodic table. Current universal MLIPs (UIPs) have been trained with the largest consistent data set available nowadays. However, these are composed mostly of bulk materials' DFT calculations. In this article, we assess the universality of all openly available UIPs, namely MACE, CHGNet, and M3GNet, in a representative task of generalization: calculation of surface energies. We find that the out-of-the-box foundation models have significant shortcomings in this task, with errors correlated to the total energy of surface simulations, having an out-of-domain distance from the training data set. Our results show that while UIPs are an efficient starting point for fine-tuning specialized models, we envision the potential of increasing the coverage of the materials space toward universal training data sets for MLIPs.

From pixels to druggable leads: A CADD strategy for the design and synthesis of potent DDR1 inhibitors.

BACKGROUND AND OBJECTIVE: While numerous in silico tools exist for target-based drug discovery, the inconsistent integration of in vitro data with predictive models hinders research and development productivity. This is particularly apparent during the Hit-to-Lead stage, where unreliable in-silico tools often lead to suboptimal lead selection. Herein, we address this challenge by presenting a CADD-guided pipeline that successfully integrates rational drug design with in-silico hits to identify a promising DDR1 lead. METHODS: 2 × 1000 ns MD simulations along with their respective FEL and MMPBSA analyses were employed to guide the rational design and synthesis of 12 novel compounds which were evaluated for their DDR inhibition. RESULTS: The molecular dynamics investigation of the initial hit led to the identification of key structural features within the DDR1 binding pocket. The identified key features were used to guide the rational design and synthesis of twelve novel derivatives. SAR analysis, biological evaluation, molecular dynamics, and free energy calculations were carried out for the synthesized derivatives to understand their mechanism of action. Compound 4c exhibited the strongest inhibition and selectivity for DDR1, with an IC50 of 0.11 µM. CONCLUSIONS: The MD simulations led to the identification of a key hydrophobic groove in the DDR1 binding pocket. The integrated approach of SAR analysis with molecular dynamics led to the identification of compound 4c as a promising lead for further development of potent and selective DDR1 inhibitors. Moreover, this work establishes a protocol for translating in silico hits to real world bioactive druggable leads.

RepurposeDrugs: an interactive web-portal and predictive platform for repurposing mono- and combination therapies.

RepurposeDrugs (https://repurposedrugs.org/) is a comprehensive web-portal that combines a unique drug indication database with a machine learning (ML) predictor to discover new drug-indication associations for approved as well as investigational mono and combination therapies. The platform provides detailed information on treatment status, disease indications and clinical trials across 25 indication categories, including neoplasms and cardiovascular conditions. The current version comprises 4314 compounds (approved, terminated or investigational) and 161 drug combinations linked to 1756 indications/conditions, totaling 28 148 drug-disease pairs. By leveraging data on both approved and failed indications, RepurposeDrugs provides ML-based predictions for the approval potential of new drug-disease indications, both for mono- and combinatorial therapies, demonstrating high predictive accuracy in cross-validation. The validity of the ML predictor is validated through a number of real-world case studies, demonstrating its predictive power to accurately identify repurposing candidates with a high likelihood of future approval. To our knowledge, RepurposeDrugs web-portal is the first integrative database and ML-based predictor for interactive exploration and prediction of both single-drug and combination approval likelihood across indications. Given its broad coverage of indication areas and therapeutic options, we expect it accelerates many future drug repurposing projects.