Cardiac fibrosis is a pivotal pathological process driving heart failure following ischemia-reperfusion (I/R) injury. STC1 (stanniocalcin-1), a conserved glycoprotein with mitochondrial regulatory functions, remains unexplored in myocardial fibrosis. This study investigates the mechanistic role of cardiomyocyte-derived STC1 in cardiac fibrosis. Mouse models of myocardial I/R injury were established by left anterior descending coronary artery ligation followed by 24-hour reperfusion. Cardiac-specific STC1 knockdown (adeno-associated virus 9-shSTC1) or overexpression (adeno-associated virus 9-STC1) preceded I/R injury induction. Cardiac function was assessed via echocardiography, and fibrosis was quantified histologically (Masson's trichrome, Sirius red) and molecularly (α-SMA [alpha smooth muscle actin], COL1A1 [collagen type I alpha 1 chain], COL3A1 [collagen type III alpha 1 chain], Smad2 [SMAD family member 2]). In vitro, cardiomyocyte-derived conditioned medium was applied to TGF-β (transforming growth factor beta)-stimulated fibroblasts, supplemented with recombinant STC1 for rescue assays. Protein-protein interactions were analyzed via coimmunoprecipitation and domain-deletion mutants. STC1 deficiency exacerbated post-I/R cardiac dysfunction and fibrosis, accompanied by elevated α-SMA, COL1A1, and phosphorylated-Smad2 levels. Conversely, STC1 overexpression attenuated I/R-induced functional decline and fibrosis. In addition, cardiomyocyte-derived STC1 suppressed TGF-β-driven fibroblast activation and Smad2 nuclear translocation. Mechanistically, STC1 directly interacted with the DOCKER domain of DOCK1 (dedicator of cytokinesis 1), disrupting DOCK1/RAC1-GTP (Rac family small GTPase 1 bound to guanosine triphosphate) signaling. Genetic ablation of the DOCKER domain abolished STC1's antifibrotic effects, restoring RAC1-GTP activity and fibrosis markers. Cardiomyocyte-derived STC1 mitigates cardiac fibrosis post I/R by targeting the DOCK1/RAC1 axis via its interaction with the DOCKER domain. These findings identify STC1-DOCK1 as a novel therapeutic pathway for ischemic heart failure.
Molecular docking aims to identify a biologically relevant binding pose of a ligand in the active site of a receptor. Reliable preparation of docking systems remains a key bottleneck in structure-based drug discovery, as it requires extensive preprocessing of receptor and ligand structures and careful configuration of docking parameters. To simplify this process and promote reproducibility through online docking experiments, we have developed CHARMM-GUI Ligand Docker, integrating four popular docking engines (AutoDock Vina, Smina, RxDock, and DiffDock) into a unified and intuitive framework for both experts and nonexperts. Users can upload and modify receptor and ligand structures, define binding sites and flexible residues, select docking programs, and execute docking calculations. The resulting docking poses can be optionally filtered using PoseBuster to remove physically implausible or geometrically strained ligand poses. Selected poses can then be seamlessly transferred to the CHARMM-GUI High-Throughput Simulator to generate molecular dynamics simulation systems and inputs for further refinement or free-energy evaluation. Ligand Docker thus provides a robust and automated platform that bridges the gap between ligand docking and simulation, ensuring that reproducible and simulation-ready systems can be prepared rapidly and conveniently for a wide range of applications in drug discovery. Ligand Docker is expected to serve as a valuable web-based resource that simplifies and accelerates the multistate docking setup.
Covalent inhibitor research is an emerging topic in drug discovery due to its superior performance in specificity and inhibition effects. While molecular docking is a popular strategy in prediction and assessment of ligand conformations or poses in receptor proteins, covalent ligand docking requires nontrivial preparation efforts, as the ligand structure changes during the covalent complex formation. In order to facilitate molecular docking for covalent ligands, we have developed CHARMM-GUI Covalent Ligand Docker (CGUI-CLD), a new module for covalent ligand docking supported by AutoDock4. CGUI-CLD automates ligand preparation, supports ligand modification, implements docking simulation, and presents results through an intuitive user interface. A knowledge-based library built in CGUI-CLD currently supports 66 warheads and 8 amino acids, which can be used to automate the covalent ligand transformation from a pre-reaction to a post-reaction adduct form seamlessly. Moreover, CHARMM-GUI High-Throughput Simulator is integrated for rapid generation of multiple molecular dynamics simulation systems. CGUI-CLD is expected to significantly reduce a massive workload of covalent ligand docking and advance covalent ligand research.
Avocado (Persea americana) is an economically important fruit with growing global production. While multiple genome assemblies and transcriptomic datasets are publicly available-including the dedicated platform AvoBase-these resources remain fragmented and lack integration with pre‑computed multi‑omics analyses and reproducible workflows. We present the Persea americana Genome Database (PAGD; http://bioinfor.kib.ac.cn/ ), an integrated platform consolidating two high‑quality genome assemblies (Hass chromosome‑level and West Indian telomere‑to‑telomere). It also hosts RNA‑seq datasets from 13 NCBI BioProjects, each linked to detailed biosample metadata, all uniformly re‑processed. Pre‑computed results include gene family classification (68 TPS genes), collinearity, gene density, and expression profiles. PAGD offers BLAST, JBrowse, interactive heatmaps, and data download. Additionally, we developed three Docker‑encapsulated Snakemake workflows for reference‑based and reference‑free transcriptome analysis, eliminating manual software configuration. PAGD advances existing avocado genomic resources by integrating multi‑omics data with pre‑computed analyses and a reproducible transcriptome workflow. The encapsulated workflows lower the technical barrier for RNA‑seq analysis, are adaptable to other plant species, and support functional genomics, breeding, and comparative studies.
Clustering analysis is a foundational step in exploratory data analysis workflows, with dimensionality reduction methods commonly used to visualize multidimensional data in lower-dimensional spaces and infer sample clustering. Principal Component Analysis (PCA) is widely applied in metabolomics but is often suboptimal for clustering visualization. Metabolomics data often require specialized manipulations such as blank removal, quality control adjustments, and data transformations that demand efficient visualization tools. However, the lack of user-friendly tools for clustering without computational expertise presents a challenge for metabolomics researchers. ClusterApp addresses this gap as a web application that performs Principal Coordinate Analysis (PCoA), expanding clustering alternatives in metabolomics. Built on a QIIME 2 Docker image, it enables PCoA computation and Emperor plot visualization. The app supports data input from GNPS, GNPS2, or user-provided spreadsheets. Freely available, ClusterApp can be locally installed as a Docker image or integrated into Jupyter notebooks, offering accessibility and flexibility to diverse users. To demonstrate the data preprocessing techniques available in ClusterApp, we analyzed two Liquid Chromatography coupled to Tandem Mass Spectrometry (LC-MS/MS) metabolomics datasets: one exploring metabolomic differences in mouse tissue samples and another investigating coral life history stages. Among the dissimilarity measures available, the Bray-Curtis measure effectively highlighted key metabolomic variations and patterns across both datasets. Targeted filtering significantly enhanced data reliability by retaining biologically relevant features, 10,617 in the coral dataset and 7,341 in the mouse dataset while eliminating noise. The combination of Total Ion Current (TIC) normalization and auto-scaling improved clustering resolution, revealing distinct separations in tissue types and life stages. ClusterApp's flexible features, such as customizable blank removal and group selection, provided tailored analyses, enhancing visualization and interpretation of metabolomic profiles. ClusterApp addresses the need for accessible, dynamic tools for exploratory data analysis in metabolomics. By coupling data transformation capabilities with PCoA on multiple dissimilarity matrices, it provides a versatile solution for clustering analysis. Its web interface and Docker-based deployment offer flexibility, accommodating a wide range of use cases through graphical or programmatic interactions. ClusterApp empowers researchers to uncover meaningful patterns and relationships in metabolomics data without requiring cumbersome data manipulation or advanced bioinformatics expertise.
The rapid growth of artificial intelligence (AI) has significantly increased computing demand, intensifying the operational strain on the computing environment. While virtualization and containerization are established technologies for resource optimization, their comparative energy efficiency and environmental impact, particularly under GPU-accelerated AI workloads, are not well-quantified. This study evaluates the energy consumption and environmental impact of virtualization and containerization technologies when using Graphics Processing Units (GPUs) for AI model execution. Employing a computer vision benchmark, the performance, GPU resource utilization, and power consumption were measured. The experiment involved training a DenseNet-121 model on the MNIST dataset within a VirtualBox virtual machine and a Docker container environment. The analysis indicates that containerization consistently surpasses virtualization in energy efficiency. Specifically, Docker container configuration demonstrated an approximately 21.6% reduction in total energy consumption, and a corresponding reduction in carbon dioxide (CO2) emissions compared to a VirtualBox virtual machine. Furthermore, containerization exhibited lower average and peak GPU utilization and power consumption. These findings demonstrate that containerization offers a more energy-efficient and environmentally sustainable approach than VirtualBox virtualization for the specific GPU-enabled AI workload evaluated in this study. Statistical significance testing indicates that the observed performance differentials are significant, supporting the validity of the results within the experimental scope of this work. Implementing containerization in this experimental setup may reduce energy consumption and environmental impact without compromising computational performance. Future studies should extend these analyses to larger neural network models, diverse AI workloads, and heterogeneous GPU platforms to enhance the generalizability of these findings beyond the current single-system experimental configuration.
Genome annotation files (GFF3/GTF) are the standard for storing genomic feature data, yet their flexibility often results in formatting inconsistencies that create bottlenecks for downstream bioinformatics analyses. A robust, unified framework is required to parse, standardise, and validate these files to ensure interoperability and facilitate complex comparative genomic tasks. We present AEGIS (Annotation Extraction and Genomic Integration Suite), a comprehensive toolkit designed to parse, correct, and standardise genome annotations. Beyond quality control, AEGIS provides advanced modules for flexible feature extraction (e.g., coding sequences, promoters) and comparative genomic analysis. Uniquely, it integrates multiple lines of evidence, including sequence homology, synteny, and coordinate-based lift-overs, to assess gene model correspondence and infer orthology. We demonstrate the utility of AEGIS by quantifying complex structural changes between Arabidopsis annotation versions and identifying high-confidence orthologues across diverse plant genomes. AEGIS is implemented in Python. Source code and documentation are freely available under the GPL-3 license at https://github.com/Tomsbiolab/aegis and as a Docker container at https://hub.docker.com/r/tomsbiolab/aegis. The package is also available on PyPI (pip install aegis-bio).
Functional genomics in plant biology relies on the generation, reuse, and long-term management of large numbers of plasmids produced through diverse cloning strategies. As collections expand across users and projects, laboratories face increasing challenges in organization, traceability, and preservation of construction histories. Existing cloning and sequence-design software supports plasmid design but does not address collaborative, laboratory-wide plasmid management. We developed PlasmiDB, an open-source, web-based database for managing plasmid collections in multi-user research environments. PlasmiDB provides structured storage of plasmid metadata, explicit tracking of plasmid genealogy, and traceability throughout the plasmid life cycle, including inter-laboratory exchanges. The system implements project-based access control and supports collaborative workflows involving staff, students, and core facilities. Implemented using a standard LAMP architecture and deployable via Docker, PlasmiDB is designed for extensibility without modification of the core database schema. A gene module links genetic targets to associated plasmids, primers, and CRISPR reagents, improving coherence between molecular constructs and experimental objectives. In our laboratory, PlasmiDB currently manages nearly 700 plasmids and facilitates reporting through automated declaration file generation. PlasmiDB complements existing cloning tools by providing traceability, collaboration support, and long-term data stewardship for plant molecular biology laboratories. The source code and Docker image are publicly available.
Reliable study databases require realistic data for validation and testing. Manual creation is slow, and rule-based generators often yield clinically implausible cases. We present LLM4ODM, a dockerized web system that uses large language models to generate synthetic, context-aware clinical study data directly from CDISC ODM v1.3 metadata. ODM definitions (e.g., ItemDefs) are transformed into structured prompts, processed by Gemini 1.5 Flash, and returned as JSON subject records. A validation layer ensures schema compliance, data types, branching logic, and temporal coherence. Across ten ODM datasets, including three from Heidelberg University Hospital, LLM4ODM achieved 100% schema adherence, >78% reduction in manual effort, and better clinical plausibility than rule-based methods based on structured expert review. All code, Docker images, prompts, and ODM files will be openly available.
Non-insulin-dependent diabetes mellitus (NIDDM) is a prevalent illness among individuals. This study elucidates the synthesis of pyrazolo-fused triazepine skeletons containing heterocycles and their assay-based in vitro and in silico antidiabetic activity against dual α-amylase and α-glucosidase enzymes followed by ab initio studies. The synthesis was assisted by acid-base bifunctional APTES-grafted magnetic nanocatalyst (CuFe2O4@GO-NH2), using dibenzalacetone (1a-g) and hydrazine (2a-b) as precursors, followed by the addition of isoniazid (3). The obtained heterocyclic compounds (4a-n) were corroborated using spectroscopic techniques. The ab initio structural insights were computed at the B3LYP/6-311G++(d,p) level of theory to investigate the HOMO-LUMO energy gap, chemical reactivity and chemical potential of the synthesized compounds. Molecular docking using the CDOCKER (CHARMm-based DOCKER) algorithm revealed appreciable binding interaction modes between the active sites of the receptor (PDB ID: 2QV4 and 3W37) and heterocyclic ligands. Among all the synthesized compounds, 4,4'-(1-(2,4-dinitrophenyl)-8-hydroxy-8-(pyridin-4-yl)-2,3,5,6,7,8-hexahydro-1H-pyrazolo-[1,5-d][1,2,4]-triazepine-2,5-diyl)bis(2-methoxyphenol) (4d) (IC50 = 123.79 µg mL-1) exhibited superior antidiabetic activity against α-amylase compared with the existing drug acarbose (IC50 = 171.8 µg mL-1), which is in accordance with the in silico and DFT studies. ADMET, Lipinski's Rule and TOPKAT descriptor studies were performed to assess the remarkable biocompatibility and toxicity. Therefore, the synthesized heterocycles could emerge as potent antidiabetic drugs for first-line treatment in the near future.
Root phenotypic analysis is closely related to crop yield and stress resistance. Although deep learning can improve the efficiency of root phenotype recognition, existing methods suffer from insufficient segmentation accuracy under complex soil backgrounds and focus on a single target. To address the issues of limited accuracy and operational complexity in existing root segmentation models, this paper proposes a novel wavelet-enhanced full-scale segmentation network. The WaveUNet+ model is based on U-Net3plus, replaces traditional downsampling with the Haar wavelet transform, and introduces the EMA module. The impact of the wavelet transform is validated using Grad-CAM, and HD95 is employed to evaluate the improvement in segmentation quality brought by the attention mechanism from the perspective of boundary accuracy. Transfer learning is used to improve model generalization, and the test results on diverse roots and various soils are compared. A Docker containerized root image segmentation method is designed to achieve convenient and practical operation, and the deployment feasibility of the model on edge devices is also verified. Our model effectively enhances the recognition of fine roots in soil backgrounds, leading to improvements across various metrics, achieving an Accuracy of 99.2%, while improving model accuracy with relatively low parameter count and model size. Compared with the original U-Net model, mIoU is increased by 1.52% and Recall by 2.93%. The results show that the model not only performs excellently on the original dataset but also maintains good generalization ability across different imaging modalities, crop species, and soil conditions. With Docker, users can achieve root image segmentation on their own computers without tedious program installation and environment configuration. In the future, we will attempt methods such as pruning and quantization to reduce model size, so as to better adapt to the deployment requirements of edge devices.
Genome-based bacterial taxonomy requires standardized and reproducible analytical workflows for species delineation and phylogenomic placement; however, the practical deployment of these workflows remains a significant barrier for experimental biologists and clinical scientists. Widely adopted tools such as Prokka, antiSMASH, and PhyloPhlAn underpin key steps in genome annotation, functional characterization, and phylogenomic reconstruction, but their practical deployment in routine laboratory settings, especially on Windows based systems, remains non trivial due to complex software dependencies and command line centric workflows. Existing solutions, including cloud-based platforms (e.g., Galaxy and KBase) and commercial software suites (e.g., CLC Genomics Workbench), partially alleviate these challenges but may also involve considerations related to data-privacy concerns, upload latency, storage quotas, shared computing resources, and recurring licensing costs. To address these limitations, we introduce TaxaScope, a graphical-interface-driven desktop workstation designed to support reproducible, genome-based bacterial taxonomy by integrating a curated set of community-validated tools for genome quality assessment, annotation, phylogenomic inference, genome relatedness estimation, and functional profiling within a unified local graphical user interface (GUI). By leveraging Docker- and Podman-based containerization behind a user-friendly frontend, TaxaScope provides version-locked, standardized execution environments across computing platforms without requiring manual dependency management or prior Linux expertise. We demonstrate the utility of TaxaScope through a comprehensive re-analysis of Pseudomonas putida KCTC 1751T, illustrating how standardized taxonomic workflows can be executed locally while automatically generating high-quality circular genome maps and interactive functional reports suitable for downstream interpretation and figure preparation directly from native tool outputs. Collectively, TaxaScope lowers the technical barrier to standardized and reproducible genome-based bacterial taxonomy by providing a private, locally controlled, containerized workflow that complements cloud-based and commercial infrastructures for routine taxonomic research. By providing a containerized and visualization-oriented desktop environment, TaxaScope facilitates the standardized execution of established genomic tools, thereby bridging the gap between complex bioinformatic workflows and consistent bacterial taxonomy.
BRIDGE is a Shiny-based application that provides an accessible, modular platform for individual and integrative multi-omics analysis. Using an independent SQLite database backend, it offers a local, private, and user-friendly environment that requires no prior computational expertise. The application supports proteomics, phospho-proteomics, and RNA-seq analyses through a comprehensive suite of visualization and analytical modules, together with an integrated multi-omics analysis pipeline. Built-in caching and asynchronous processing improve responsiveness, enabling efficient exploration, analysis, and visualization of multi-omics datasets on moderate hardware. BRIDGE is implemented in R using Shiny and is freely available as a Docker container at https://ghcr.io/paulilab/bridge. A public demonstration server with example datasets is available at https://bridge.imp.ac.at. Code and datasets are also available at https://github.com/paulilab/BRIDGE and under DOI: https://doi.org/10.5281/zenodo.20215824.
Biosynthetic gene clusters (BGC) are genomic regions that encode the production of specialized metabolites, including antibiotics, pigments, and toxins. While BGC are traditionally classified into broad categories such as NRPS, PKS, and terpene clusters, these classes often overlook finer relationships among gene clusters that produce structurally or functionally related compounds. Tools like BiG-SCAPE and BiG-SLiCE have been developed to address this issue by organizing BGC into gene cluster families (GCFs). CORASON complements these tools by enabling phylogenetic reconstruction of BGC, identifying conserved core genes, and visualizing GFCs as a continuum of variation in gene presence/absence and sequence identity. Although CORASON is incorporated in BiG-SCAPE visualization, it is also a standalone tool initially designed for bacterial genomes annotated via RAST and implemented through Docker in Linux environments. Here, we demonstrate CORASON's broader applicability using fungal GenBank files and its installation via Conda on Windows. As a case study, we examine metagenome-assembled genomes (MAGs) from Fusarium domesticum, a lesser-known member of the Fusarium genus, which is often present in food-associated microbiomes. Unlike its pathogenic relatives (F. oxysporum, F. graminearum), F. domesticum remains understudied, making it an interesting target for genomic mining. This work expands the accessibility of CORASON for fungal genome analysis and highlights its potential in uncovering novel biosynthetic potential in overlooked microbial taxa.
Accurate clinical interpretation of genetic variants requires integration of functional predictions, evolutionary constraint, population allele frequencies, and clinical evidence from heterogeneous resources. Conventional workflows based on standalone tools such as ensembl variant effect predictor (VEP) and ANNOVAR require complex manual configuration of plugins and databases, generate verbose transcript-level outputs unsuitable for clinical review, and rely on ad hoc scripts for format conversion and prioritization. These limitations hinder reproducibility and scalability, making data interpretation a major bottleneck in genomic medicine. We present MuSA (Multi-Source variant Annotation), an nf-core-compliant Nextflow pipeline that automates germline variant annotation from resource setup to clinical interpretation. MuSA supports both a streamlined basic mode for diagnostic workflows and an extended deep-annotation mode for comprehensive analyses. The pipeline integrates Ensembl VEP with 22 curated plugins (including AlphaMissense, CADD, SpliceAI, and Enformer), ANNOVAR, a standalone pre-configured dbNSFP distribution, the RENOVO pathogenicity predictor, and automated ACMG/AMP classification via GeneBe and InterVar. MuSA standardizes input VCFs, executes parallel annotation branches in a fully containerized workflow, and consolidates results into richly annotated mutation annotation format (MAF) files (up to 920 columns per variant), alongside interactive HTML reports tailored for clinical review with HPO-matched gene panels. Benchmarked on a WES-like dataset of 22,705 variants derived from the public GIAB NA12878/HG001 GRCh38 benchmark VCF, MuSA completes full extended-mode annotation in approximately 20 min on a 64-core server. Systematic comparison with nf-core/sarek and nf-core/variantprioritization demonstrates that MuSA uniquely combines automated resource management with YAML-based version tracking and SHA-256 integrity verification, native dbNSFP integration, RENOVO-based VUS prioritization, HPO-driven gene panel filtering, and a clinically oriented interactive HTML report; those features are mostly absent in existing nf-core annotation pipelines. Containerization through Docker/Singularity and predefined execution profiles support reproducible deployment across workstations, HPC clusters, and cloud environments. MuSA provides an end-to-end framework for clinically oriented germline variant annotation and prioritization, addressing key limitations of manual and general-purpose workflows. Its dual-output design bridges research (machine-readable MAF files compatible with downstream tools such as maftools) and clinical diagnostics (interpretation-ready HTML reports), supporting reproducible and standardized variant interpretation across teams. Current limitations include restriction to germline small variants on hg38, a substantial storage footprint (up to 223.5 GB for extended mode), and dependence on external APIs for ACMG/AMP classification and phenotype-driven filtering. The RENOVO-based VUS prioritization module is experimental and requires expert review before clinical interpretation.
Hereditary Spastic Paraplegia (HSP) is a rare neurodegenerative disorder causing progressive weakness and spasticity in the lower limbs. variants in the HPDL gene are linked to Spastic Paraplegia 83 (SPG83), an autosomal recessive form of HSP. While HPDL variants are known to cause SPG83, the molecular mechanisms behind its role remains unclear, mostly due to rare nature of the condition. The primary objective was to identify the genetic cause of HSP in two Iranian consanguineous families. Whole-exome sequencing (WES) was employed to identify genetic variants in the probands. Molegro Virtual Docker (MVD), a cutting-edge integrated platform, was utilized to perform protein-ligand docking simulations. This approach aimed to characterize the structural and functional consequences of the identified variants associated with SPG83 pathogenicity. WES identified two biallelic variants in HPDL: c.3G > C, a start-loss variant abolishing the canonical initiation codon, and c.128G > C, a missense variant. The c.128G > C variant is novel and is documented here for the first time in an SPG83 patient. Trio-based co-segregation analysis confirmed inheritance of variants. Furthermore, a comprehensive literature review revealed a significant consanguinity rate (49.55%) within families harboring HPDL variants. This study expands the genetic and clinical spectrum of HPDL variants. The identification of the genetic variants in the probands underscores the clinical value of genetic testing methods like WES as a valuable diagnostic tool.
Avoiding dual-axis visualization improves quantitative interpretation. Yet for large-scale paired datasets, combining raw data with summary metrics on dual axes often hinders interpretability, while single-axis displays risk visual saturation of paired lines. These limitations may be overcome by developing summary metrics that can be plotted on the same axis as the underlying data. We developed EZ-Pair Graph as a suite of highly scalable methods that aggregate positional and slope information of numerous lines into a unified and interpretable axis. EZ-Pair Graph comprises three complementary tools: trapezoid plot, which summarizes ascending and descending groups of paired differences and their prevalence, and the clustered line or parallel arrow plot, which can reveal clustered patterns and directional heterogeneity in paired differences. By selectively emphasizing the rank and magnitude of paired differences, these plots facilitate the interpretation of distributional differences in large-scale paired data. We demonstrate the effectiveness of our methods using biological datasets that are difficult to visualize using conventional approaches. In each case, our methods revealed structured, localized, and heterogeneous trends through clear visual summaries. As datasets increase in scale and complexity, EZ-Pair Graph may be useful for detecting underlying patterns and localized variations that are often overlooked in conventional paired-data visualizations. https://github.com/010049nn/EZ_pair_graph; EZ-Pair Graph outputs are available in multiple formats (PDF, SVG, PNG, HTML, and JSON). Installation via pip and docker is possible. Release archive DOI: 10.5281/zenodo.20437542.
This research examines the evolution of health information systems for recording medical incidents at sea within the framework of Telemedical Maritime Assistance Services (T.M.A.S.). The article provides a comprehensive comparative analysis between a low-code implementation and two new architectures: a monolithic approach (Flask/SQLite) and a decoupled approach oriented towards microservices (React/Node.js/ Docker). It also attempts to compare the capabilities of Large Language Models (LLMs), which have evolved from being less-than-ideal code-writing assistants to becoming efficient Principal Full-Stack Engineers. The findings demonstrate the exponential leap in the capabilities of Generative Artificial Intelligence and highlight that Monolithic Flask is the ideal solution for a small research lab or a land-based hospital with a fast network, as its maintenance time is minimal. However, for the shipping industry, the React/Node.js architecture is undoubtedly superior.
Cardiotoxicity remains a critical concern in drug development, often leading to late-stage attrition of promising compounds. While traditional assessments focus on Kv11.1 channel inhibition, the Comprehensive in Vitro Proarrhythmic Assay (CiPA) initiative emphasizes the importance of evaluating additional cardiac ion channels, notably Cav1.2 and Nav1.5. In this study, we address the limitations of existing machine learning (ML) models, which typically rely on Kv11.1-specific data, by developing a deep learning (DL) framework that integrates inhibition data across all three key ion channels. A large and diverse data set (Cardio-Tox) was curated by combining experimental data from the PubChem, CUPID, and CToxPred2 repositories, totaling 34,124 molecules for Kv11.1, 1564 for Cav1.2, and 3217 for Nav1.5. Using this data set, trained GNN models are capable of individual channel prediction. The developed CardiotoxPred method, which includes the Kv, Cav, and Nav models, achieved an average prediction accuracy of 86.7% on a test data set. In addition to robust predictive performance, GNNExplainer offers interpretable visualizations by highlighting atom- and bond-level contributions via colors. These insights support cardiac molecular severity estimation, optimization, and safety profiling. All the models are freely accessible via GitHub in a user-friendly Docker container, providing a practical tool for early-stage cardiotoxicity risk assessment in drug discovery pipelines.
Genetic diagnosis of endocrine disorders is often hampered by the complexity of analyzing Whole Exome Sequencing (WES) data. We developed the endocrine variant extractor (EVE), a streamlined, clinician-friendly bioinformatics pipeline designed for multi-tier genetic screening with a core panel for parathyroid disorders (26 genes) and an expanded endocrine panel for broader metabolic assessment (413 genes, fully encompassing the parathyroid panel). Encapsulated within a Docker container and automated via a custom Python wrapper, EVE integrates core bioinformatics engines, including BWA-MEM, GATK, and SnpEff. The pipeline employs a tiered reporting strategy, filtering and annotating variants across both panels using pathogenicity scores (SIFT, PolyPhen-2) and clinical databases (ClinVar, gnomAD). This architecture ensures cross-platform compatibility without complex manual configuration. To validate the pipeline, EVE was applied to clinical datasets. EVE successfully filtered >300,000 raw variants down to a handful of actionable candidates. Using this pipeline, we precisely identified the first Korean case of a de novo GATA3 frameshift variant (p.Ala173fs) in an HDR syndrome patient, which was not previously reported in the ClinVar database. Analysis took ~3 h, reducing manual data review by >99.6%. EVE provides a streamlined, high-efficiency workflow that automates the filtering of thousands of raw WES variants into a curated list of clinically relevant variants. This robust framework enables the creation of a comprehensive "endocrine variant atlas," empowering clinicians to integrate high-throughput genetic profiling into routine diagnostics and accelerate the discovery of novel disease-causing variants. The complete source code for EVE is freely available at https://github.com/hanyunseo01/EVE.