Adaptive malware detection using sequential feature selection: A dueling double deep Q-Network framework for intelligent classification

We present PDBTools.jl, a lightweight and high-performance Julia package for reading, writing, selecting, and analyzing molecular structure data stored in PDB and mmCIF (PDBx) file formats. The package provides a compact and memory-efficient atom representation based on inline strings and single-precision floating-point coordinates, enabling the handling of very large structures on standard hardware. A flexible and customizable atom selection syntax inspired by VMD is augmented by native support for arbitrary Julia functions as selectors, offering exceptional expressiveness and performance for dynamic queries. Beyond file I/O and selection, PDBTools.jl includes high-performance implementations of key structural analysis algorithms: solvent-accessible surface area (SASA) calculation via the Shrake-Rupley method with Fibonacci lattice sampling, hydrogen bond detection, contact and distance maps, backbone dihedral angles for Ramachandran analysis, secondary structure assignment through integration with STRIDE and DSSP, and protein transfer free energy (m-value) calculations using the Tanford additive model. Cell list-based neighbor finding ensures O(N) scaling for distance-dependent operations with support for periodic boundary conditions. PDBTools.jl is designed for molecular dynamics simulation workflows and integrates with Chemfiles.jl, MolSimToolkit.jl, and ComplexMixtures.jl. The package is freely available under the MIT license from the Julia General Registry, and full documentation can be found at https://m3g.github.io/PDBTools.jl.

PlasticAnalytics provides an automated workflow that addresses key bottlenecks in vibrational spectroscopic analysis of microplastics by Raman spectroscopy and Fourier transform infrared spectroscopy (FTIR). The preprocessing framework integrates an iterative asymmetric penalized least-squares (i-arPLS) baseline correction algorithm optimized for spectra with complex environmental backgrounds, coupled with a hybrid rule-based and machine learning framework that automatically removes spurious peaks (cosmic rays and CO2) while handling resampling, normalization, and smoothing. A complementary machine learning module identifies and removes substrate spectra in spectral images, ensuring that downstream classification operates only on particulate-derived signals. The pipeline combines these steps with a deep residual network and an uncertainty-aware quality-control classifier trained on virgin, consumer, and environmentally weathered plastic spectra, achieving classification accuracies of 96.9% (Raman) and 97.9% (FTIR) and matching or exceeding existing architectures. For spectral imaging, automated background removal and high-speed inference reduced processing time by over 90%, from more than 200 min (Raman) and 800 min (FTIR) to under 7 min in both cases. PlasticAnalytics supports the major instrument platforms and file formats, providing a scalable, reproducible pipeline for environmental microplastic analysis.

Many single-cell transcriptomics datasets are stored in the HDF5-backed AnnData (H5AD) file format, as popularised by the Python scverse ecosystem. However, accessing these datasets from R, allowing users to take advantage of the strengths of each language, can be difficult. anndataR facilitates this access by allowing users to natively read and write H5AD files in R, convert them to and from SingleCellExperiment or Seurat objects, or even work with the resulting R AnnData object directly. We perform rigorous testing to ensure compatibility between Python-written and R-written H5AD files, guaranteeing long-term interoperability between languages. anndataR's source code is available on GitHub at scverse/anndataR under the MIT license. It is compatible with R version 4.5, has been archived at 10.5281/zenodo.18775712 and included within Bioconductor: 10.18129/B9.bioc.anndataR. Installation instructions and tutorials can be found in the online documentation at anndatar.scverse.org. Issues can be reported at the GitHub repository. Code to reproduce the analyses performed can be found on GitHub at LouiseDck/anndataR-paper, archived at 10.5281/zenodo.18792241.

Elucidating protein-ligand interactions is pivotal for understanding biological mechanisms and accelerating drug discovery. Blind docking, which identifies binding sites without prior knowledge, has become an indispensable computational strategy for analyzing the surge of protein structures generated by Cryo-EM and AI-based prediction tools like AlphaFold3. Our previous server, CB-Dock2, has been widely adopted by the global research community, averaging over 1000 daily submissions since July 2022 due to its accuracy and user-friendliness. Building on this foundation and incorporating extensive user feedback, we present CB-Dock3, a substantially enhanced platform. Key upgrades include a refined docking engine, an expanded template library, and support for diverse file formats. Benchmark evaluations on CASF-2016 demonstrate that CB-Dock3 achieves a success rate of 67.4% (RMSD ≤ 2.0 Å), representing a 10.6 percentage-point absolute improvement over its predecessor and outperforming other popular blind docking tools. Additionally, CB-Dock3 introduces critical new features driven by community needs: support for user-defined docking regions to handle large complexes, and a metal-aware protocol that explicitly retains essential metal ions and cofactors during simulation. CB-Dock3 stands as an accurate, rapid, and accessible resource for the scientific community, freely available at https://cadd.labshare.cn/cb-dock3/.

Ethnic music, as a valuable treasure created in the long-term historical development of the Chinese nation, has the unique cultural elements of the Chinese nation in its musical form, artistic value and ideological concepts, and its integration into the music curriculum of preschool education in colleges and universities can help students to deepen their understanding of the ethnic culture and promote the inheritance of ethnic music culture. For this reason, the study builds an immersive situation in the ethnic music classroom of preschool education in colleges and universities, and explores the effective path of the music generation model in promoting the reform of experiential ethnic music teaching. Based on the elaboration of related music knowledge and MIDI file format, we innovatively constructed a multi-track ethnomusicological music generation model Tr-MTMG based on Transformer, which consists of three parts: data preprocessing network, learning network and generating network, and is capable of generating multi-track music. Through a series of experiments, in the Lakh MIDI dataset, the entropy value of the level histogram of this model is 2.8504, which is closest to the real samples, and it proves that the model based on this paper can generate more harmonized and more realistic music. The music created by this model achieved a high score of 105.65 in the Turing test. After multiple evaluations, it is verified that the model proposed in this paper can generate good quality music, and can be effectively applied to music tasks in pre-school education in universities.

To examine the potential errors of a general large language model (LLM) (ie, Claude 3.5 Sonnet) on data extraction from randomised controlled trials (RCTs). An empirical study comparing Claude 3.5 Sonnet extractions against a human-performed verification dataset. The extraction tasks for Claude 3.5 Sonnet were based solely on original RCT portable document format (PDF) files. For PDFs that could not be directly extracted by Claude 3.5 Sonnet, optical character recognition was employed to convert them into text format before extraction. A random sample of 664 trials was selected from a well-established trial bank and a final data pool was established based on rigorous manual cross-checking as a reference standard. PubMed, EMBASE, Scopus, Web of Science (all databases) and the Cochrane Central Register of Controlled Trials (CENTRAL) up to February 2023. RCTs on children involving medication and adverse events. Claude 3.5 Sonnet was applied to extract the basic information (eg, trial design, population information and source of funding) and adverse outcomes (ie, name of adverse events, number of events). Claude 3.5 Sonnet outputs were compared against the final data pool and all errors were recorded. Results are presented as error rates and with 95% CI, estimated using a generalised linear mixed model. For the 664 trials, a total of 23 069 data cells were extracted via Claude 3.5 Sonnet, with 10 624 for basic information and 12 445 for adverse outcomes. The overall error rate for data extraction was 6.6% (95% CI 5.4% to 8.2%), with 5.7% (95% CI 5.2% to 6.1%) in basic information and 7.6% (95% CI 4.9% to 11.8%) in adverse outcomes. When stratified the 1542 total errors by error types, misallocation (assigning data to incorrect fields; 57.1%, 881/1542) and missed or omitted data (incomplete extraction of available data; 23.2%, 357/1542) accounted for the two most frequent errors, with misallocation occurring more in basic information (53.3%, 470/881), while missed or omitted data occurred more in adverse outcomes (96.1%, 343/357). Post hoc analysis examining the association between trial reporting quality (assessed using Consolidated Standards of Reporting Trials (CONSORT) 2025 and LLM data extraction error rates indicated that higher CONSORT adherence was associated with lower extraction error rates. The data extraction error of Claude was relatively low, but it alerts LLM applications in evidence synthesis. Detailed checking for LLM outputs should be the primary consideration for evidence synthesisers.

Serologic typing for ABO and RhD is standard in transfusion services, with extended serology and genotyping performed to reduce red cell alloimmunization risk. In Kenya, RH typing is limited to RhD, and genotyping is unavailable. This study used RHD/RHCE genotyping to predict phenotypes and their distribution in a Kenyan blood donor population. A total of 191 donors (114 D-, 74 D+, and 3 weak D) from the Kenya National Blood Transfusion Service were selected. Next-generation sequencing was performed on DNA extracts using a targeted blood group sequencing panel (Illumina MiSeq). Variant call format (VCF) files were annotated with wANNOVAR, and phenotypes were predicted by matching VCF data to the International Society of Blood Transfusion (ISBT) Blood Group database. RHD*01N.01, RHD*08N.01 (RHD*Ψ), and RHD*03N.01 alleles were identified predicting D- phenotype. Discordant phenotype results were observed in 11 samples with genotype predicting nine D+ (partial D) in 114 D-, one D- in 74 D+, and one D- in three weak D phenotypes. For RHCE, 15 allele types produced 30 genotypes with 63% carrying at least one RHCE variant allele linked to: 1) weak and/or partial c and e, 2) hrB-, and 3) V+/-, VS+, phenotypes. Genotyping revealed RhD/RHD phenotype/genotype discrepancies and RH allele diversity among Kenyan donors, including RHCE variants affecting high- and low-prevalence antigen expression. These findings highlight the role of genotyping to improve accuracy for RH typing to minimize the risk of patient alloimmunization.

This study aimed to evaluate and compare the performance of five AI algorithms for tooth segmentation and labeling on intraoral scans,as their performance remains unclear. A total of 100 intraoral scans in the STL file format were classified into two main groups: complete dentition (C) and partial dentition (P, fewer than 12 teeth). Each group was further divided by arch into four subgroups (n = 25 each): complete maxillary (Mx-C), complete mandibular (Md-C), partial maxillary (Mx-P), and partial mandibular (Md-P). The algorithms tested were the Tooth Group Network (Team CGIP), Dentbird Studio (Dentbird), Medit Ortho Simulation (Medit), NemoSmile 3D (Nemotec), and MovumStudio (MovumTech). Manual segmentations by an expert operator served as the ground truth. Performance was assessed using Python and five metrics, with Intersection over Union (IOU) as the primary indicator. Statistical analysis included permutation tests with the Bonferroni-Holm correction (α = 0.05). Significant differences were observed between groups and algorithms (P<0.05). IOU scores ranged from 0.72 to 0.92 in complete dentition and showed greater variability in partial dentition (0-0.928). The Tooth Group Network and MovumStudio consistently outperformed the others, with MovumStudio achieving the highest performance across all metrics and groups. Its performance matched that of a human expert when compared against a subset of the data. Tooth segmentation and labeling performance vary depending on dentition completeness and algorithm choice. MovumStudio demonstrated the most robust and consistent results, comparable to expert human annotation.

This study attempted to compare frictional resistance among three-dimensional (3D) printed resin brackets, and also with conventional metal and ceramic brackets under in vitro conditions. 3D bracket designs were created digitally and converted to stereolithography (STL) file format. Subsequently, brackets were produced using a FormLabs 3B+ printer using Permanent Crown and Biomed Clear resins. Half of these 3D-printed brackets were aged by thermal cycling, and then the frictional resistances of aged resin, non-aged resin, ceramic and metal brackets were measured using a universal testing machine. Shrinkage was 5.41% for Permanent Crown brackets and 7.84% for Biomed Clear brackets. Resin-based brackets produced by 3D printing demonstrated the highest frictional resistance. After the aging process, the frictional resistance of the 3D-printed brackets was reduced. Frictional resistance was lowest for the stainless steel wire and stainless steel bracket combination, and highest for the beta-titanium wire and resin bracket combinations (P < 0.05). As the Permanent Crown resin bracket demonstrated frictional resistance comparable to ceramic brackets after aging, advances in more durable 3D printing materials with improved surface properties are expected to further enhance the clinical applicability of 3D-printed brackets.

The reusability of proteomics data sets depends on the ability to obtain accurate metadata to guide reprocessing pipelines. However, many data sets deposited in public data repositories lack sufficient and reliable annotation, limiting large-scale reanalyses. To address this challenge, we developed RunAssessor, a tool that systematically extracts and summarizes information directly from mass spectrometry data files prior to peptide identification analysis. RunAssessor extracts and summarizes sample preparation and instrument acquisition parameters directly from the data where possible. Using one complete data set and test files from 18 other data sets as examples, we demonstrate RunAssessor's ability to extract instrument models, isobaric labels, phosphoenrichment, precursor and fragment ion tolerances, along with the dynamic exclusion time used by the instrument. These extracted metadata are stored in a comprehensive output file, and summarized in a standard Sample and Data Relationship Format (SDRF) file, thereby reducing the burden of manual curation and improving the reliability of proteomics data set metadata, facilitating the reuse of public data.

GEFF: Graph Exchange File Format

ABSTRACT Effective CAD data management is crucial for engineering design productivity, yet challenges in file organization, interoperability, and workflow efficiency persist. While prior studies addressed technical aspects of PDM systems and file formats, the interaction between technical solutions and human factors remains underexplored. This study bridges this gap by examining the combined impact of structured training and file format standardization on CAD workflow performance, evaluating 120 professionals in control and experimental groups and assessing improvements in archiving, project categorization, security practices, and employing a QFD analysis to assess 35 CAD formats through expert testing, establishing performance hierarchies for manufacturing, AEC, and media applications. Key findings revealed a significant improvement after training, with large effect sizes across variables, reflecting meaningful enhancements in consistency, file retrieval, storage management, and security practices. Furthermore, based on expert ratings and normalized weighting, this study revealed a Pareto distribution where dwg, dxf, IGES, STEP, STL, and SVG collectively accounted for the majority of the weighted TIR, guiding format selection for specific industry applications. Manufacturing workflows benefited most from IGES and STEP formats, whereas AEC collaboration prioritized DWG and DXF. Media and entertainment pipelines relied on SVG, OBJ, and mb/.ma formats for texture and animation efficiency. These results show that human competency and technical interoperability in CAD data management must be developed together. By highlighting practical improvements in human–technical interactions and providing data-driven format hierarchies, the study offers actionable insights for workflow optimization, cross-platform collaboration, and adoption of interoperable CAD formats in professional settings, enhancing efficiency and standardized data management.

Combinatorial barcoding technologies for single-cell nucleotide sequencing, such as split-pool ligation protocols, involve sequential rounds of cell barcoding to uniquely tag individual cells. The rapid adoption of combinatorial barcoding in recent years is due in part to its scalability across cells and samples. However, small shifts in barcode positions within sequencing reads caused by technical artifacts, e.g. during barcode incorporation or synthesis, can impact the accurate assignment of reads to cell barcodes. Existing processing tools typically assume barcodes contain fixed-length nucleotide sequences located at fixed positions within reads, overlooking any positional variability. Consequently, reads containing truncated or mispositioned barcodes are discarded during initial data processing steps leading to significant data loss. To solve this limitation and maximize the retention of sequencing reads from single-cell combinatorial barcoding experiments, we introduce scarecrow. This tool screens a subsample of reads to generate position-specific barcode profiles, which are then used to flexibly identify barcode sequences in each read whilst accounting for positional errors, a phenomenon we refer to as "jitter". Barcode matches are then prioritized to minimize nucleotide mismatches and the degree of jitter. These initial profiles are subsequently used to extract and error correct barcode combinations in high throughput sequencing libraries. By incorporating jitter into barcode error correction, scarecrow enables greater data recovery and improved downstream single-cell analyses. Scarecrow is fully open access, implemented in Python, and generates output files using standardized sequence file formats for maximal interoperability. A detailed explanation of the scarecrow workflow can be found in the supplementary materials. Scarecrow is freely available on GitHub https://github.com/MorganResearchLab/scarecrow and Zenodo https://doi.org/10.5281/zenodo.18621784.

Defect phase diagrams provide a unified description of crystal defect states for materials design and are central to the scientific objectives of the Collaborative Research Centre (CRC) 1394. Their construction requires the systematic integration of heterogeneous experimental and simulation data across research groups and locations. In this setting, research data management (RDM) is a key enabler of new scientific insight by linking distributed research activities and making complex data reproducible and reusable. To address the challenge of heterogeneous data sources and formats, a comprehensive RDM infrastructure has been established that links experiment, data and analysis in a seamless workflow. The system combines: (1) a joint electronic laboratory notebook and laboratory information management system, (2) easy‐to‐use large‐object data storage, (3) automatic metadata extraction from heterogeneous and proprietary file formats, (4) interactive provenance graphs for data exploration and reuse and (5) automated reporting and analysis workflows. The two key technological elements are the openBIS electronic laboratory notebook and laboratory information management system, and a newly developed companion application that extends openBIS with large‐scale data handling, automated metadata capture and federated access to distributed research data. This integrated approach reduces friction in data capture and curation, enabling traceable and reusable datasets that accelerate the construction of defect phase diagrams across institutions.

This study investigates the extraction and analysis of metadata embedded in image and video files, focusing on the differences between iOS and Android platforms across various file formats. Metadata was extracted using three specialized tools: Metadata2go, ExifMeta, and ExifInfo. The research involved systematically retrieving files from both platforms, organizing them by format and operating system and conducting a comparative evaluation. Eight key metadata attributes were analyzed: device make, model, software version, timestamp, geographic coordinates (longitude and latitude), color space and lens model. Results indicate that iOS devices consistently retain more detailed and structured metadata than Android devices, reflecting architectural differences in metadata handling. This study underscores the critical role of metadata in digital forensics and information management.

Raman spectroscopy offers a uniquely rich window into molecular structure and composition, making it a powerful tool across fields ranging from materials science to biology. However, the reproducibility of Raman data analysis remains a fundamental bottleneck. In practice, transforming raw spectra into meaningful results is far from standardized: workflows are often complex, fragmented, and implemented through highly customized, case-specific code. This challenge is compounded by the lack of unified open-source pipelines and the diversity of acquisition systems, each introducing its own file formats, calibration schemes, and correction requirements. Consequently, researchers must frequently rely on manual, ad hoc reconciliation of processing steps. To address this gap, we introduceTRaP(Toolbox for Reproducible Raman Processing), an open-source, GUI-based Python toolkit designed to bring reproducibility, transparency, and portability to Raman spectral analysis. TRaP unifies the entire preprocessing-to-analysis pipeline within a single, coherent framework that operates consistently across heterogeneous instrument platforms (e.g., Cart, Portable, Renishaw, and MANTIS). Central to its design is the concept of fully shareable, declarative workflows: users can encode complete processing pipelines into a single configuration file (e.g., JSON), enabling others to reproduce results instantly without reimplementing code or reverse-engineering undocumented steps. Beyond convenience, TRaP integrates configuration management, X-axis calibration, spectral response correction, interactive processing, and batch execution into a workflow-driven architecture that enforces deterministic, repeatable operations. Every transformation is explicitly recorded, making the full processing history transparent, inspectable, and reproducible. This eliminates ambiguity in how results are generated and ensures that identical protocols can be applied consistently across datasets and experimental contexts. Through representative use cases, we show that TRaP enables seamless, reproducible preprocessing of Raman spectra acquired from diverse platforms within a unified environment. We hope TRaP can empower Raman data processing as a reproducible, shareable, and systematized scientific practice, aligning it with modern standards for computational research. TRaP is released as an open-source software athttps://github.com/hrlblab/TRaP

There has been a marked increase in use of a noninvasive functional imaging technique called optoretinography (ORG). As more groups use ORGs, it is crucial to have a consistent methodology, and understand what analysis parameters influence repeatability. In this work, we present an open-source software library called ƒ(Cell) designed to facilitate reproducible and repeatable analyses of ORG data. We designed ƒ(Cell) as a Python software library that can co-register and analyze ORG datasets, while also enabling process auditing. To validate the software, we used our previously obtained normative optoretinography datasets as well as datasets from six other groups. We performed a variance decomposition analysis of all ƒ(Cell) parameters. Using ƒ(Cell), we successfully read and analyzed all seven datasets, despite varying data structures, file formats, and imaging protocols. Our full factorial analysis of 12,960 parameter permutations across 16 normative individuals resulted in 3 significant parameters (P < 0.001) with at least a small effect (ƞ2 > 0.01): segmentation radius, normalization, and relativization method, with ƞ2 = 0.213, 0.037, and 0.012, respectively. In a full factorial analysis of analysis parameters, we found that the mean subtraction relativization method, a segmentation radius matched to the spacing of the cells of interest, and score-based normalization resulted in the lowest coefficient of variance. Overall, we found that ƒ(Cell) enables consistent, reproducible, and auditable analyses of intensity-based ORG (iORG) datasets. This work is designed to facilitate reproducible analyses of ORG datasets to aid in the use of ORG in clinical trials.

The accurate measurement of spheroid area and morphology is critical for the progression of the integration of 3D models in in vitro cancer research and is increasingly used to measure effective therapeutic efficacy of X-ray radiation. Current methods of measuring spheroids require labour-intensive manual analysis or the use of complex software tools. SphereMetrics was created as a user-friendly Shiny app with a straightforward interface designed to streamline the process of measuring the area and eccentricity of spheroids. It allows the upload and automated detection of spheroids across multiple file formats and generates robust and objective area and eccentricity measurements. Area measurements derived from SphereMetrics were compared to manual quantification with ImageJ and AnaSP for untreated and irradiated (0–20 Gy) human neuroendocrine BON-1 cancer spheroids. When compared to ImageJ and AnaSP, SphereMetrics was shown to provide fast, accurate data (R2 = 0.87 and 0.83, respectively). Spheroid analysis took 19.92 ± 8 s/image with SphereMetrics, approximately four times faster than ImageJ analysis (89.81 ± 11.52 s/image) and nine times faster than AnaSP (183.36 ± 31.62 s/image). SphereMetrics represents an accessible and efficient tool for spheroid analysis, facilitating data collection and analysis for routine in vitro model research, ideal for non-programmers.

Hydrogen/deuterium exchange-mass spectrometry (HX-MS) is a rapidly expanding technique used to investigate protein conformational ensembles. The growing popularity and utility of HX-MS has driven the development of diverse instrumentation and software, resulting in inconsistent, non-standardized data analysis and representation. Most HX-MS data formats also employ only mean deuteration representations of the data rather than full isotopic mass spectra, which reduces the information content of the data and limits downstream quantitative analysis. Inspired by reliable protein structure and genomics data formats, we present HXMS, a unified, lightweight, scalable, and human-readable file format for HX-MS data. The HXMS format preserves the isotopic mass envelopes for all peptides, captures the full experimental time-course including fully deuterated control samples, and contains all other key information. It supports multimodal distributions, post-translational modifications (PTMs), and experimental replicates. To promote compatibility with existing HX-MS workflows, we also developed PFLink, a Python package that converts exported data files from commonly used HX-MS software to the HXMS format. PFLink and the HXMS format will enable quantitative, higher-resolution data processing, improved data sharing and storage among HX-MS practitioners, future machine learning applications, and further developments in HX-MS analysis. PFLink is publicly available to install locally on HuggingFace, alongside documentation, or use online at HuggingFace (https://huggingface.co/spaces/glasgow-lab/PFlink). The supplementary information includes sample input files, sample HXMS files, and a generic unfilled PFlink custom CSV file that users may populate with key experimental conditions and results, which can then be read and converted into the HXMS format.