Discovery Methods Research Articles

Advances in Deep Learning Assisted Drug Discovery Methods: A Self-review

: Artificial Intelligence is a field within computer science that endeavors to replicate the intricate structures and operational mechanisms inherent in the human brain. Machine learning is a subfield of artificial intelligence that focuses on developing models by analyzing training data. Deep learning is a distinct subfield within artificial intelligence, characterized by using models that depict geometric transformations across multiple layers. The deep learning has shown significant promise in various domains, including health and life sciences. In recent times, deep learning has demonstrated successful applications in drug discovery. In this self-review, we present recent methods developed with the aid of deep learning. The objective is to give a brief overview of the present cutting-edge advancements in drug discovery from our group. We have systematically discussed experimental evidence and proof of concept examples for the deep learning-based models developed, such as Deep- BindBC, DeepPep, and DeepBindRG. These developments not only shed light on the existing challenges but also emphasize the achievements and prospects for future drug discovery and development progress.

Syntactically branching out beyond the traditional classroom: A report on the discovery method: Supplementary material

Syntactically branching out beyond the traditional classroom: A report on the discovery method: Supplementary material

Syntactically branching out beyond the traditional classroom: A report on the discovery method

Abstract: This article reports on the successful implementation of the discovery method in an undergraduate Generative Syntax course. Rather than reading textbooks on syntactic analysis and doing exercises to apply what was read, students embark on a nine-stage learning cycle consisting of interactive, guided, and independent work. They engage with problem sets to build their own analyses of syntactic phenomena step by step. This method—meant to resemble the work of real-life scientists—allows students to hypothesize, experiment, and build knowledge from trial and error (Kolb 1984, Prince 2004). To measure the effectiveness of our pedagogy, we analyzed course evaluations submitted from 2012 to the present. Learner feedback shows that they find the assignments and activities helpful, that the course challenges them to think and learn, and that they consider the weekly write-ups demanding, yet highly beneficial for their learning.

DNA Barcoding of the Endemic Freshwater Catfish Encheloclarias tapeinopterus (Bleeker, 1853) (Siluriformes:Clariidae) from Bangka Island, Indonesia

Abstract Encheloclarias tapeinopterus, an endemic catfish exclusively found in Bangka Island, Indonesia. The population of this species has been observed to be decreasing at an alarming rate, which has resulted in the IUCN Red List classifying it as a species that is vulnerable (VU). During our expedition on the 25th and 26th of January 2023, a total of Two specimens of E. tapeinopterus were collected from Kelumpik River, Pinang Sebatang Village, Central Bangka Regency, Bangka Island, Indonesia. In this groundbreaking study, we proudly present the pioneering documentation of E. tapeinopterus DNA barcoding, focusing on the Cytochrome C oxidase Subunit I (COI) gene. The recorded DNA Barcode of E. tapeinopterus has been officially registered in the Genbank under the accession code OQ281707. By employing DNA barcoding, we offer an advanced and expedient method for precise species identification and discovery, revolutionizing the field of taxonomy. Phylogenetic analysis revealed that E. tapeinopterus forms a distinct clade alongside the Clarias and Heterobranchus genus, shedding light on its evolutionary relationships and genetic affinities.

Molecular Docking and In Silico ADME(T) Evaluation of Selective Phytochemical Inhibitors of VEGF2 Target for the Treatment of Diabetic Retinopathy

Background: Diabetic retinopathy (DR) is the leading cause of vision loss in diabetic patients. Currently, the treatment involves the use of glucocorticoids or a VEGF antagonist, which are "off-label" at present. However, the conventional method of drug discovery and development is a time-consuming process that requires more than a decade of meticulous research and huge financial support. While there are a few effective small organic molecules against DR that were identified many years ago, nutraceuticals - naturally available functional foods containing vitamins, antioxidants, minerals, fatty acids, and amino acids - can also help delay the progression of some diseases. Methods: In this study, 43 phytochemical constituents from four medicinal plants were tested for their binding affinity to the influential VEGFR2 target of diabetic retinopathy. The study used a computational approach, in silico molecular docking study, structure-based drug design approach, MSD (Molecular Dynamic Simulation analysis), in silico ADME(T) studies. Results: The study reported that all phytochemical constituents displayed good to the highest binding affinity than the standard ruboxistaurin. Six phytochemical constituents, namely terchebulin, pedunculagin, punicalagin, punicalin, casuariniane, and chebulagic acid, exhibited equipotent to higher activity than the standard. These constituents displayed conventional hydrogen bonds, pialkyl, and pi-cation interactions to achieve their high binding affinity. The highest binding scores were chosen for analysis using MSD, ensuring stability of the ligand-protein complex. Pharmacodynamic and pharmacokinetic properties were evaluated, and their safety profile was validated. Conclusion: This in silico screening study suggests that active phytomolecules present in medicinal plants may inhibit the VEGFR2 target. The best-docked compounds, possessing drug-like properties, can be used to develop potential inhibitors against DR or to mitigate its severity.

A hybrid constrained continuous optimization approach for optimal causal discovery from biological data.

Understanding causal effects is a fundamental goal of science and underpins our ability to make accurate predictions in unseen settings and conditions. While direct experimentation is the gold standard for measuring and validating causal effects, the field of causal graph theory offers a tantalizing alternative: extracting causal insights from observational data. Theoretical analysis has shown that this is indeed possible, given a large dataset and if certain conditions are met. However, biological datasets, frequently, do not meet such requirements but evaluation of causal discovery algorithms is typically performed on synthetic datasets, which they meet all requirements. Thus, real-life datasets are needed, in which the causal truth is reasonably known. In this work we first construct such a large-scale real-life dataset and then we perform on it a comprehensive benchmarking of various causal discovery methods. We find that the PC algorithm is particularly accurate at estimating causal structure, including the causal direction which is critical for biological applicability. However, PC does only produces cause-effect directionality, but not estimates of causal effects. We propose PC-NOTEARS (PCnt), a hybrid solution, which includes the PC output as an additional constraint inside the NOTEARS optimization. This approach combines PC algorithm's strengths in graph structure prediction with the NOTEARS continuous optimization to estimate causal effects accurately. PCnt achieved best aggregate performance across all structural and effect size metrics. https://github.com/zhu-yh1/PC-NOTEARS.

Stop-Probability-Based Network Topology Discovery Method

Stop-Probability-Based Network Topology Discovery Method

Development of an in vitro compound screening system that replicate the in vivo spine phenotype of idiopathic ASD model mice.

Autism Spectrum Disorder (ASD) is a developmental condition characterized by core symptoms including social difficulties, repetitive behaviors, and sensory abnormalities. Aberrant morphology of dendritic spines within the cortex has been documented in genetic disorders associated with ASD and ASD-like traits. We hypothesized that compounds that ameliorate abnormalities in spine dynamics might have the potential to ameliorate core symptoms of ASD. Because the morphology of the spine is influenced by signal inputs from other neurons and various molecular interactions, conventional single-molecule targeted drug discovery methods may not suffice in identifying compounds capable of ameliorating spine morphology abnormalities. In this study, we focused on spine phenotypes in the cortex using BTBR T + Itpr3 tf /J (BTBR) mice, which have been used as a model for idiopathic ASD in various studies. We established an in vitro compound screening system using primary cultured neurons from BTBR mice to faithfully represent the spine phenotype. The compound library mainly comprised substances with known target molecules and established safety profiles, including those approved or validated through human safety studies. Following screening of this specialized library containing 181 compounds, we identified 15 confirmed hit compounds. The molecular targets of these hit compounds were largely focused on the 5-hydroxytryptamine receptor (5-HTR). Furthermore, both 5-HT1AR agonist and 5-HT3R antagonist were common functional profiles in hit compounds. Vortioxetine, possessing dual attributes as a 5-HT1AR agonist and 5-HT3R antagonist, was administered to BTBR mice once daily for a period of 7days. This intervention not only ameliorated their spine phenotype but also alleviated their social behavior abnormality. These results of vortioxetine supports the usefulness of a spine phenotype-based assay system as a potent drug discovery platform targeting ASD core symptoms.

Fused in silico and bioactivity evaluation method for drug discovery: T001-10027877 was identified as an antiproliferative agent that targets EGFRT790M/C797S/L858R and EGFRT790M/L858R

BackgroundFacing the significant challenge of overcoming drug resistance in cancer treatment, particularly resistance caused by mutations in epidermal growth factor receptor (EGFR), the aim of our study was to identify potent EGFR inhibitors effective against the T790M/C797S/L858R mutant, a key player in resistance mechanisms.MethodsOur integrated in silico approach harnessed machine learning, virtual screening, and activity evaluation techniques to screen 5105 compounds from three libraries, aiming to find candidates capable of overcoming the resistance conferred by the T790M and C797S mutations within EGFR. This methodical process narrowed the search down to six promising compounds for further examination.ResultsKinase assays identified three compounds to which the T790M/C797S/L858R mutant exhibited increased sensitivity compared to the T790M/L858R mutant, highlighting the potential efficacy of these compounds against resistance mechanisms. Among them, T001-10027877 exhibited dual inhibitory effects, with IC50 values of 4.34 µM against EGFRT790M/C797S/L858R and 1.27 µM against EGFRT790M/L858R. Further investigations into the antiproliferative effects in H1975, A549, H460 and Ba/F3-EGFRL858/T790M/C797S cancer cells revealed that T001-10027877 was the most potent anticancer agent among the tested compounds. Additionally, the induction of H1975 cell apoptosis and cell cycle arrest by T001-10027877 were confirmed, elucidating its mechanism of action.ConclusionsThis study highlights the efficacy of combining computational techniques with bioactivity assessments in the quest for novel antiproliferative agents targeting complex EGFR mutations. In particular, T001-10027877 has great potential for overcoming EGFR-mediated resistance and merits further in vivo exploration. Our findings contribute valuable insights into the development of next-generation anticancer therapies, demonstrating the power of an integrated drug discovery approach.

Harnessing Deep Learning Methods for Voltage-Gated Ion Channel Drug Discovery.

Voltage-gated ion channels (VGICs) are pivotal in regulating electrical activity in excitable cells and are critical pharmaceutical targets for treating many diseases including cardiac arrhythmia and neuropathic pain. Despite their significance, challenges such as achieving target selectivity persist in VGIC drug development. Recent progress in deep learning, particularly diffusion models, has enabled the computational design of protein binders for any clinically relevant protein based solely on its structure. These developments coincide with a surge in experimental structural data for VGICs, providing a rich foundation for computational design efforts. This review explores the recent advancements in computational protein design using deep learning and diffusion methods, focusing on their application in designing protein binders to modulate VGIC activity. We discuss the potential use of these methods to computationally design protein binders targeting different regions of VGICs, including the pore domain, voltage-sensing domains, and interface with auxiliary subunits. We provide a comprehensive overview of the different design scenarios, discuss key structural considerations, and address the practical challenges in developing VGIC-targeting protein binders. By exploring these innovative computational methods, we aim to provide a framework for developing novel strategies that could significantly advance VGIC pharmacology and lead to the discovery of effective and safe therapeutics.

Machine Learning-Driven Discovery and Evaluation of Antimicrobial Peptides from Crassostrea gigas Mucus Proteome.

Marine antimicrobial peptides (AMPs) represent a promising source for combating infections, especially against antibiotic-resistant pathogens and traditionally challenging infections. However, traditional drug discovery methods face challenges such as time-consuming processes and high costs. Therefore, leveraging machine learning techniques to expedite the discovery of marine AMPs holds significant promise. Our study applies machine learning to develop marine AMPs, focusing on Crassostrea gigas mucus rich in antimicrobial components. We conducted proteome sequencing of C. gigas mucous proteins, used the iAMPCN model for peptide activity prediction, and evaluated the antimicrobial, hemolytic, and cytotoxic capabilities of six peptides. Proteomic analysis identified 4490 proteins, yielding about 43,000 peptides (8-50 amino acids). Peptide ranking based on length, hydrophobicity, and charge assessed antimicrobial potential, predicting 23 biological activities. Six peptides, distinguished by their high relative scores and promising biological activities, were chosen for bactericidal assay. Peptides P1 to P4 showed antimicrobial activity against E. coli, with P2 and P4 being particularly effective. All peptides inhibited S. aureus growth. P2 and P4 also exhibited significant anti-V. parahaemolyticus effects, while P1 and P3 were non-cytotoxic to HEK293T cells at detectable concentrations. Minimal hemolytic activity was observed for all peptides even at high concentrations. This study highlights the potent antimicrobial properties of naturally occurring oyster mucus peptides, emphasizing their low cytotoxicity and lack of hemolytic effects. Machine learning accurately predicted biological activity, showcasing its potential in peptide drug discovery.

IoT-Driven Healthcare an Ensemble Learning Approach for Early Detection and Prevention of Chronic Diseases

This investigate presents an IoT-driven healthcare system leveraging outfit learning methods for early discovery and avoidance of unremitting illnesses. By joining information from wearable sensors, shrewd domestic gadgets, and electronic wellbeing records, the system offers a comprehensive wellbeing checking framework. Gathering learning calculations, counting Random Forest, Slope Boosting, Stacking, and Bolster Vector Machines, are utilized to combine numerous models' qualities, improving forecast exactness and robustness. Experimental comes about on an assorted dataset illustrate the viability of the proposed approach, accomplishing a precision of 85%, accuracy of 87%, review of 84%, F1-score of 85%, and AUC-ROC of 0.92. Comparative examinations with standard calculations and related work within the field emphasize the prevalence of the outfit learning approach in leveraging IoT information for healthcare applications. This investigation contributes to progressing IoT-driven healthcare by giving bits of knowledge into the potential of gathering learning procedures for illness expectation and avoidance. By making strides early location and personalized mediations, the system points to upgraded understanding results, diminishing healthcare costs, and progress in populace well-being management.

DIS-Guard: Enhancing SDN resilience to topology and RCO attacks

SDN controllers employ discovery protocols with LLDP packets to discover the network topology; however, this approach is susceptible to topology attacks due to the static reuse of LLDP packets without proper origin verification or integrity checks. This paper proposes a novel form of topology attack called the Round-trip time Confusion Onslaught (RCO) attack. Our findings showcase the capability of the RCO attack to mislead the discovery service of SDN controllers and bypass the Real-time Link Verification (RLV) defence mechanism, which depends on timing features in Machine Learning (ML) models. The RCO attack significantly reduces the effectiveness of the RLV mechanism in terms of True Positive Rate (TPR), Precision, F1-score, and Cohen’s Kappa and increases the False Positive Rate (FPR). These results underscore the difficulties in achieving dependable ML accuracy in topology attack detection. Accordingly, we propose the DIS-Guard framework, designed to protect and improve the efficiency of discovery protocols within SDN controllers. This framework utilizes the mutual Transport Layer Security (mTLS) protocol to discover hosts proactively. Moreover, it employs a novel combination of a chaos-tent map and the proposed statistical method for link discovery. Through extensive evaluation, we illustrate DIS-Guard’s superior performance and security over the established Ryu (switches.py) controller and the RLV defence mechanism. Notably, DIS-Guard reduces host-to-host access delays, decreases LLDP packet transmission and reception significantly, and improves control channel efficiency. Collectively, these quantifiable benefits manifest the efficacy of DIS-Guard as a significantly optimized solution for SDN security and performance.

An end-to-end deep learning method for mass spectrometry data analysis to reveal disease-specific metabolic profiles

Untargeted metabolomic analysis using mass spectrometry provides comprehensive metabolic profiling, but its medical application faces challenges of complex data processing, high inter-batch variability, and unidentified metabolites. Here, we present DeepMSProfiler, an explainable deep-learning-based method, enabling end-to-end analysis on raw metabolic signals with output of high accuracy and reliability. Using cross-hospital 859 human serum samples from lung adenocarcinoma, benign lung nodules, and healthy individuals, DeepMSProfiler successfully differentiates the metabolomic profiles of different groups (AUC 0.99) and detects early-stage lung adenocarcinoma (accuracy 0.961). Model flow and ablation experiments demonstrate that DeepMSProfiler overcomes inter-hospital variability and effects of unknown metabolites signals. Our ensemble strategy removes background-category phenomena in multi-classification deep-learning models, and the novel interpretability enables direct access to disease-related metabolite-protein networks. Further applying to lipid metabolomic data unveils correlations of important metabolites and proteins. Overall, DeepMSProfiler offers a straightforward and reliable method for disease diagnosis and mechanism discovery, enhancing its broad applicability.

Enhancing novel isoform discovery: leveraging nanopore long-read sequencing and machine learning approaches.

Long-read sequencing technologies can capture entire RNA transcripts in a single sequencing read, reducing the ambiguity in constructing and quantifying transcript models in comparison to more common and earlier methods, such as short-read sequencing. Recent improvements in the accuracy of long-read sequencing technologies have expanded the scope for novel splice isoform detection and have also enabled a far more accurate reconstruction of complex splicing patterns and transcriptomes. Additionally, the incorporation and advancements of machine learning and deep learning algorithms in bioinformatic software have significantly improved the reliability of long-read sequencing transcriptomic studies. However, there is a lack of consensus on what bioinformatic tools and pipelines produce the most precise and consistent results. Thus, this review aims to discuss and compare the performance of available methods for novel isoform discovery with long-read sequencing technologies, with 25 tools being presented. Furthermore, this review intends to demonstrate the need for developing standard analytical pipelines, tools, and transcript model conventions for novel isoform discovery and transcriptomic studies.

Machine Learning Guided AQFEP: A Fast and Efficient Absolute Free Energy Perturbation Solution for Virtual Screening.

Structure-based methods in drug discovery have become an integral part of the modern drug discovery process. The power of virtual screening lies in its ability to rapidly and cost-effectively explore enormous chemical spaces to select promising ligands for further experimental investigation. Relative free energy perturbation (RFEP) and similar methods are the gold standard for binding affinity prediction in drug discovery hit-to-lead and lead optimization phases, but have high computational cost and the requirement of a structural analog with a known activity. Without a reference molecule requirement, absolute FEP (AFEP) has, in theory, better accuracy for hit ID, but in practice, the slow throughput is not compatible with VS, where fast docking and unreliable scoring functions are still the standard. Here, we present an integrated workflow to virtually screen large and diverse chemical libraries efficiently, combining active learning with a physics-based scoring function based on a fast absolute free energy perturbation method. We validated the performance of the approach in the ranking of structurally related ligands, virtual screening hit rate enrichment, and active learning chemical space exploration; disclosing the largest reported collection of free energy simulations to date.

Physics and chemistry from parsimonious representations: image analysis via invariant variational autoencoders

Electron, optical, and scanning probe microscopy methods are generating ever increasing volume of image data containing information on atomic and mesoscale structures and functionalities. This necessitates the development of the machine learning methods for discovery of physical and chemical phenomena from the data, such as manifestations of symmetry breaking phenomena in electron and scanning tunneling microscopy images, or variability of the nanoparticles. Variational autoencoders (VAEs) are emerging as a powerful paradigm for the unsupervised data analysis, allowing to disentangle the factors of variability and discover optimal parsimonious representation. Here, we summarize recent developments in VAEs, covering the basic principles and intuition behind the VAEs. The invariant VAEs are introduced as an approach to accommodate scale and translation invariances present in imaging data and separate known factors of variations from the ones to be discovered. We further describe the opportunities enabled by the control over VAE architecture, including conditional, semi-supervised, and joint VAEs. Several case studies of VAE applications for toy models and experimental datasets in Scanning Transmission Electron Microscopy are discussed, emphasizing the deep connection between VAE and basic physical principles. Python codes and datasets discussed in this article are available at https://github.com/saimani5/VAE-tutorials and can be used by researchers as an application guide when applying these to their own datasets.

Software application profile: tpc and micd-R packages for causal discovery with incomplete cohort data.

The Peter Clark (PC) algorithm is a popular causal discovery method to learn causal graphs in a data-driven way. Until recently, existing PC algorithm implementations in R had important limitations regarding missing values, temporal structure or mixed measurement scales (categorical/continuous), which are all common features of cohort data. The new R packages presented here, micd and tpc, fill these gaps. micd and tpc packages are R packages. The micd package provides add-on functionality for dealing with missing values to the existing pcalg R package, including methods for multiple imputations relying on the Missing At Random assumption. Also, micd allows for mixed measurement scales assuming conditional Gaussianity. The tpc package efficiently exploits temporal information in a way that results in a more informative output that is less prone to statistical errors. The tpc and micd packages are freely available on the Comprehensive R Archive Network (CRAN). Their source code is also available on GitHub (https://github.com/bips-hb/micd; https://github.com/bips-hb/tpc).

Innovative Alignment-Based Method for Antiviral Peptide Prediction.

Antiviral peptides (AVPs) represent a promising strategy for addressing the global challenges of viral infections and their growing resistances to traditional drugs. Lab-based AVP discovery methods are resource-intensive, highlighting the need for efficient computational alternatives. In this study, we developed five non-trained but supervised multi-query similarity search models (MQSSMs) integrated into the StarPep toolbox. Rigorous testing and validation across diverse AVP datasets confirmed the models' robustness and reliability. The top-performing model, M13+, demonstrated impressive results, with an accuracy of 0.969 and a Matthew's correlation coefficient of 0.71. To assess their competitiveness, the top five models were benchmarked against 14 publicly available machine-learning and deep-learning AVP predictors. The MQSSMs outperformed these predictors, highlighting their efficiency in terms of resource demand and public accessibility. Another significant achievement of this study is the creation of the most comprehensive dataset of antiviral sequences to date. In general, these results suggest that MQSSMs are promissory tools to develop good alignment-based models that can be successfully applied in the screening of large datasets for new AVP discovery.

New combined Inverse-QSAR and molecular docking method for scaffold-based drug discovery

Computer-aided drug discovery plays a vital role in developing novel medications for various diseases. The COVID-19 pandemic has heightened the need for innovative approaches to design lead compounds with the potential to become effective drugs. Specifically, designing promising inhibitors of the SARS-CoV-2 main protease (Mpro) is crucial, as it plays a key role in viral replication. Phytochemicals, primarily flavonoids and flavonols from medicinal plants, were screened. Fifty small molecules were selected for molecular docking analysis against SARS-CoV-2 Mpro (PDB ID: 6LU7). Binding energies and interactions were analyzed and compared to those of the anti-SARS-CoV-2 inhibitor Nirmatrelvir. Using these 50 structures as a training set, a QSAR model was built employing simple, reversible topological descriptors. An inverse-QSAR analysis was then performed on 2⁹ = 512 hydroxyl combinations at nine possible positions on the flavone and flavonol scaffold. The model predicted three novel, promising compounds exhibiting the most favorable binding energies (−8.5 kcal/mol) among the 512 possible hydroxyl combinations: 3,6,7,2′,4′-pentahydroxyflavone (PF9), 6,7,2′,4′-tetrahydroxyflavone (PF11), and 3,6,7,4′-tetrahydroxyflavone (PF15). Molecular dynamics (MD) simulations demonstrated the stability of the PF9/Mpro complex over 300 ns of simulation. These predicted structures, reported here for the first time, warrant synthesis and further evaluation of their biological activity through in vitro and in vivo studies.