Source Code Research Articles

SurfR: Riding the wave of RNA-seq data with a comprehensive bioconductor package to identify surface protein-coding genes.

Proteins at the cell surface connect signaling networks and largely determine a cell's capacity to communicate and interact with its environment. In particular, variations in transcriptomic profiles are often observed between healthy and diseased cells, leading to distinct sets of cell-surface proteins. For these reasons, cell-surface proteins may act as biomarkers for the detection of cells of interest in tissues or body fluids, are often the target of pharmaceutical agents, and hold significant promise in the clinical practice for diagnosis, prognosis, treatment development, and evaluation of therapy response. Therefore, implementing robust methods to identify condition-specific cell-surface proteins is of pivotal importance to advance biomedical research. We developed SurfR, an R/Bioconductor package providing a streamlined end-to-end workflow for computationally identifying surface protein-coding genes from expression data. Our user-friendly, comprehensive workflow performs systematic expression data retrieval from public databases, differential gene expression across conditions, integration of datasets, enrichment analysis, identification of targetable proteins on a condition of interest, and data visualization. SurfR is released under GNU-GPL-v3.0 License. Source code, documentation, examples, and tutorials are available through Bioconductor (http://www.bioconductor.org/packages/SurfR). RMD notebooks with the use cases code described in the manuscript can be found on GitHub (https://github.com/auroramaurizio/SurfR_UseCases).

AutoTurb: Using large language models for automatic algebraic turbulence model discovery

Symbolic regression (SR) methods have been extensively investigated to explore explicit algebraic Reynolds stress models (EARSM) for turbulence closure of Reynolds-averaged Navier-Stokes (RANS) equations. The deduced EARSM can be readily implemented in existing computational fluid dynamic (CFD) codes and promotes the identification of physically interpretable turbulence models. Recently, large language models (LLMs) trained on large amounts of publicly available source code have drawn great attention for their abilities to automatically discover equations with more general free-text inputs and problem descriptions and provide wider possibilities with novel insights. This work proposes a novel framework, named “AutoTurb,” using LLMs to automatically discover algebraic expressions for correcting the linear Reynolds stress model. The direct Reynolds stress and the indirect RANS output (e.g., velocity field) are both involved in the training objective to guarantee data consistency and avoid numerical stiffness. An evolutionary search framework is used for global optimization, where constraints on functional complexity and simulation convergence are integrated into the objective to manage the extensive flexibility of LLMs. The proposed method is performed for separated flow over periodic hills. The generalizability of the discovered model is verified on a set of 2D turbulent separated flows with different Reynolds numbers and geometries. Results show that the corrected RANS enhances predictions of both Reynolds stress and mean velocity fields. Compared to models from other studies, our discovered model shows superior accuracy and generalization capability. The proposed approach provides a promising paradigm for using LLMs to improve turbulence modeling for a given class of flows.

ACTION: Augmentation and computation toolbox for brain network analysis with functional MRI.

Functional magnetic resonance imaging (fMRI) has been increasingly employed to investigate functional brain activity. Many fMRI-related software/toolboxes have been developed, providing specialized algorithms for fMRI analysis. However, existing toolboxes seldom consider fMRI data augmentation, which is quite useful, especially in studies with limited or imbalanced data. Moreover, current studies usually focus on analyzing fMRI using conventional machine learning models that rely on human-engineered fMRI features, without investigating deep learning models that can automatically learn data-driven fMRI representations. In this work, we develop an open-source toolbox, called Augmentation and Computation Toolbox for braIn netwOrk aNalysis (ACTION), offering comprehensive functions to streamline fMRI analysis. The ACTION is a Python-based and cross-platform toolbox with graphical user-friendly interfaces. It enables automatic fMRI augmentation, covering blood-oxygen-level-dependent (BOLD) signal augmentation and brain network augmentation. Many popular methods for brain network construction and network feature extraction are included. In particular, it supports constructing deep learning models, which leverage large-scale auxiliary unlabeled data (3,800+ resting-state fMRI scans) for model pretraining to enhance model performance for downstream tasks. To facilitate multi-site fMRI studies, it is also equipped with several popular federated learning strategies. Furthermore, it enables users to design and test custom algorithms through scripting, greatly improving its utility and extensibility. We demonstrate the effectiveness and user-friendliness of ACTION on real fMRI data and present the experimental results. The software, along with its source code and manual, can be accessed online.

NURBS curve shape prior-guided multiscale attention network for automatic segmentation of the inferior alveolar nerve.

Accurate segmentation of the inferior alveolar nerve (IAN) within Cone-Beam Computed Tomography (CBCT) images is critical for the precise planning of oral and maxillofacial surgeries, especially to avoid IAN damage. Existing methods often fail due to the low contrast of the IAN and the presence of artifacts, which can cause segmentation discontinuities. To address these challenges, this paper proposes a novel approach that employs Non-Uniform Rational B-Spline (NURBS) curve shape priors into a multiscale attention network for the automatic segmentation of the IAN. Firstly, an automatic method for generating non-uniform rational B-spline (NURBS) shape prior is proposed and introduced into the segmentation network, which significantly enhancing the continuity and accuracy of IAN segmentation. Then a multiscale attention segmentation network, incorporating a dilation selective attention module is developed, to improve the network's feature extraction capacity. The proposed approach is validated on both in-house and public datasets, showcasing superior performance compared to established benchmarks, achieving 80.29±11.04% dice coefficient (Dice) and 68.14±12.06% intersection of union (IoU), the 95% Hausdorff distance (95HD) reaches 1.61±6.14 mm and mean surface distance (MSD) reaches 0.64±2.16 mm on private dataset. On public dataset, the Dice reaches 80.69±4.93%, IoU reaches 67.86±6.73%, 95HD reaches 1.04±0.95 mm, and MSD reaches 0.42±0.34 mm. Compared to state-of-the-art networks, the proposed approach out-performs in both voxel accuracy and surface distance. It offers significant potential to improve doctors' efficiency in segmentation tasks and holds promise for applications in dental surgery planning. The source codes are available at https://github.com/SJTUjsl/NURBS_IAN.git.

Plastic waste identification based on multimodal feature selection and cross-modal Swin Transformer

The classification and recycling of municipal solid waste (MSW) are strategies for resource conservation and pollution prevention, with plastic waste identification being an essential component of waste sorting. Multimodal detection of solid waste has increasingly replaced single-modal methods constrained by limited informational capacity. However, existing hyperspectral feature selection algorithms and multimodal identification methods have yet to leverage cross-modal information exhaustively. Therefore, two RGB-hyperspectral image (RGB-HSI) multimodal instance segmentation datasets were constructed to support research in plastic waste sorting. A feature band selection algorithm based on the Activation Weight function was proposed to automatically select influential hyperspectral bands from multimodal data, thereby reducing the burden of data acquisition, transmission, and inference. Furthermore, the multimodal Selective Feature Network (SFNet) was introduced to balance information across various modalities and stages. Moreover, the Correlation Swin Transformer Block was proposed, specifically crafted to fuse cross-modal mutual information, which can be synergistically employed with SFNet to enhance multimodal recognition capabilities further. Experimental results show that the Activation Weight band selection function can select the most effective feature bands. At the same time, the Correlation SF-Swin Transformer achieved the highest F1-scores of 97.85% and 97.37% in the two plastic waste object detection experiments, respectively. The source code and final models are available at https://github.com/Bazenr/Correlation-SFSwin, and the dataset can be accessed at https://www.kaggle.com/datasets/bazenr/rgb-hsi-rgb-nir-municipal-solid-waste.

A step towards quantifying, modeling and exploring uncertainty in biomedical knowledge graphs

Objective:This study aims at automatically quantifying and modelling the uncertainty of facts in biomedical knowledge graphs (BKGs) based on their textual supporting evidence using deep learning techniques. Materials and Methods:A sentence transformer is employed to extract deep features of sentences used to classify sentence factuality using a naive Bayes classifier. For each fact and its supporting evidence in a source KG, the deep feature extractor and the classifier are used to quantify the factuality of each sentence which are then transformed to numerical values in [0,1] before being averaged to get the confidence score of the fact. Results:The fact classification feature extractor enhances the separability of classes in the embedding space. This helped the fact classification model to achieve a better performance than existing factuality classification with hand-crafted features. Uncertainty quantification and modelling were demonstrated on SemMedDB by creating USemMedDB, showing KGB2U’s ability to process large BKGs. A subset of USemMedDB facts is modelled to demonstrate the correlation between the structure of the uncertain BKG and the confidence scores. The best-trained model is used to predict confidence scores of existing and unseen facts. The top-ranked unseen facts were grounded using scientific evidence showing KGB2U’s ability to discover new knowledge. Conclusion:Supporting literature of BKG facts can be used to automatically quantify their uncertainty. Additionally, the resulting uncertain biomedical KGs can be used for knowledge discovery. BKG2U interface and source code are available at http://biofunk.datanets.org/ and https://github.com/BahajAdil/KBG2U respectively.

GraCEImpute: A novel graph clustering autoencoder approach for imputation of single-cell RNA-seq data.

Single-cell RNA sequencing (scRNA-seq) technology establishes a unique view for elucidating cellular heterogeneity in various biological systems. Yet the scRNA-seq data is compromised by a high dropout rate due to the technological limitation, and the substantial data loss poses computational challenges on subsequent analyses. This study introduces a novel graph clustering autoencoder (GCAE)-based imputation approach (GraCEImpute) to address the challenge of missing data in scRNA-seq data. Our comprehensive evaluation demonstrates that the GraCEImpute model outperforms existing approaches in accurately imputing dropout zeros within scRNA-seq data. The proposed GraCEImpute model also demonstrates the significantly enhanced quality of downstream scRNA-seq data analyses, including clustering, differential gene expression (DEG) analysis, and cell trajectory inference. These improvements underscore the GraCEImpute model's potential to facilitate a deeper understanding of cellular processes and heterogeneity through the scRNA-seq data analyses. The source code is released at https://www.healthinformaticslab.org/supp/.

Medical Federated Model with Mixture of Personalized and Shared Components.

Although data-driven methods usually have noticeable performance on disease diagnosis and treatment, they are suspected of leakage of privacy due to collecting data for model training. Recently, federated learning provides a secure and trustable alternative to collaboratively train model without any exchange of medical data among multiple institutes. Therefore, it has draw much attention due to its natural merit on privacy protection. However, when heterogenous medical data exists between different hospitals, federated learning usually has to face with degradation of performance. In the paper, we propose a new personalized framework of federated learning to handle the problem. It successfully yields personalized models based on awareness of similarity between local data, and achieves better tradeoff between generalization and personalization than existing methods. After that, we further design a differentially sparse regularizer to improve communication efficiency during procedure of model training. Additionally, we propose an effective method to reduce the computational cost, which improves computation efficiency significantly. Furthermore, we collect five real medical datasets, including two public medical image datasets and three private multi-center clinical diagnosis datasets, and evaluate its performance by conducting nodule classification, tumor segmentation, and clinical risk prediction tasks. Comparing with 14 existing related methods, the proposed method successfully achieves the best model performance, and meanwhile up to 60% improvement of communication efficiency. Source code is public, and can be accessed at: https://github.com/ ApplicationTechnologyOfMedicalBigData/pFedNet-code.

Evaluating and extending speedup techniques for optimal crossing minimization in layered graph drawings.

A layered graph is an important category of graph in which every node is assigned to a layer, and layers are drawn as parallel or radial lines. They are commonly used to display temporal data or hierarchical graphs. Previous research has demonstrated that minimizing edge crossings is the most important criterion to consider when looking to improve the readability of such graphs. While heuristic approaches exist for crossing minimization, we are interested in optimal approaches to the problem that prioritize human readability over computational scalability. We aim to improve the usefulness and applicability of such optimal methods by understanding and improving their scalability to larger graphs. This paper categorizes and evaluates the state-of-the-art linear programming formulations for exact crossing minimization and describes nine new and existing techniques that could plausibly accelerate the optimization algorithm. Through a computational evaluation, we explore each technique's effect on calculation time and how the techniques assist or inhibit one another, allowing researchers and practitioners to adapt them to the characteristics of their graphs. Our best-performing techniques yielded a median improvement of 2.5-17× depending on the solver used, giving us the capability to create optimal layouts faster and for larger graphs. We provide an open-source implementation of our methodology in Python, where users can pick which combination of techniques to enable according to their use case. A free copy of this paper and all supplemental materials, datasets used, and source code are available at https://osf.io/5vq79.

Efficient federated unlearning under plausible deniability

Privacy regulations like the GDPR in Europe and the CCPA in the US allow users the right to remove their data from machine learning (ML) applications. Machine unlearning addresses this by modifying the ML parameters in order to forget the influence of a specific data point on its weights. Recent literature has highlighted that the contribution from data point(s) can be forged with some other data points in the dataset with probability close to one. This allows a server to falsely claim unlearning without actually modifying the model’s parameters. However, in distributed paradigms such as federated learning (FL), where the server lacks access to the dataset and the number of clients are limited, claiming unlearning in such cases becomes a challenge. An honest server must modify the model parameters in order to unlearn. This paper introduces an efficient way to achieve machine unlearning in FL, i.e., federated unlearning, by employing a privacy model which allows the FL server to plausibly deny the client’s participation in the training up to a certain extent. Specifically, we demonstrate that the server can generate a Proof-of-Deniability, where each aggregated update can be associated with at least x (the plausible deniability parameter) client updates. This enables the server to plausibly deny a client’s participation. However, in the event of frequent unlearning requests, the server is required to adopt an unlearning strategy and, accordingly, update its model parameters. We also perturb the client updates in a cluster in order to avoid inference from an honest but curious server. We show that the global model satisfies (ϵ,δ)-differential privacy after T number of communication rounds. The proposed methodology has been evaluated on multiple datasets in different privacy settings. The experimental results show that our framework achieves comparable utility while providing a significant reduction in terms of memory (≈ 30 times), as well as retraining time (1.6-500769 times). The source code for the paper is available https://github.com/Ayush-Umu/Federated-Unlearning-under-Plausible-Deniability.

Tight Exponential Strong Converse for Source Coding Problem with Encoded Side Information

Tight Exponential Strong Converse for Source Coding Problem with Encoded Side Information

GridapTopOpt.jl: a scalable Julia toolbox for level set-based topology optimisation

In this paper, we present GridapTopOpt, an extendable framework for level set-based topology optimisation that can be readily distributed across a personal computer or high-performance computing cluster. The package is written in Julia and uses the Gridap package ecosystem for parallel finite element assembly from arbitrary weak formulations of partial differential equations (PDEs) along with the scalable solvers from the Portable and Extendable Toolkit for Scientific Computing (PETSc). The resulting user interface is intuitive and easy-to-use, allowing for the implementation of a wide range of topology optimisation problems with a syntax that is near one-to-one with the mathematical notation. Furthermore, we implement automatic differentiation to help mitigate the bottleneck associated with the analytic derivation of sensitivities for complex problems. GridapTopOpt is capable of solving a range of benchmark and research topology optimisation problems with large numbers of degrees of freedom. This educational article demonstrates the usability and versatility of the package by describing the formulation and step-by-step implementation of several distinct topology optimisation problems. The driver scripts for these problems are provided and the package source code is available at https://github.com/zjwegert/GridapTopOpt.jl.

CVE VULNERABILITY CLASSIFICATION IN SOURCE CODE BASED ON TOKEN ANALYSIS AND LSTM NETWORKS

As web applications become increasingly widespread, the importance of source code security is growing rapidly. Exposed vulnerabilities present serious risks to both service providers and customers. Various models have been proposed to address this issue; however, most approaches rely on complex graph structures generated from source code or on expert-driven regular expression patterns. This paper introduces a model that utilizes token-based mechanisms combined with deep learning techniques for efficient vulnerability detection in PHP (Hypertext Preprocessor) web applications. By leveraging the PHP tokenization process, we have developed a custom token that merges tokens, supports key PHP features, and optimizes parsing. Using datasets such as the Software Assurance Reference Dataset (SARD) and SQLI-LABS, this paper demonstrates the training of a deep learning model with enhanced tokens to effectively detect vulnerabilities in the source code.

Global heterogeneous graph convolutional network: from coarse to refined land cover and land use segmentation

ABSTRACT The abundant details embedded in very-high-resolution remote sensing images establish a solid foundation for comprehending the land surface. Simultaneously, as spatial resolution advances, there is a corresponding escalation in the required granularity of land cover and land use (LCLU) categories. The coarse classes identified necessitate further refinement into more detailed categories. For instance, the ‘built-up’ class can be subdivided into specific categories such as squares, stadiums, and airports. These refined LCLU classifications are better equipped to support diverse domains. Nonetheless, most studies simply adopt methods initially designed for coarse LCLU when addressing the challenging refined LCLU segmentation. Few studies have considered the inherent relationships between coarse and refined LCLU, overlooking the potential exploitation of the numerous recently released LCLU products. To better leverage this prior knowledge, we propose the Global Heterogeneous Graph Convolutional Network (GHGCN). The GHGCN introduces a heterogeneous graph and excels in establishing relationships between coarse and refined LCLU, which can extract long-distance dependencies more effectively than convolutional neural networks. Furthermore, this model is performed end-to-end, eliminating the necessity for presegmentation and facilitating training acceleration. GHGCN exhibits competitive performance compared to state-of-the-art models, indicating its effective design in exploiting coarse LCLU data, especially for categories with limited samples. The source code is released at: https://github.com/Liuzhizhiooo/GHGCN.

NovoRank: Refinement for De Novo Peptide Sequencing Based on Spectral Clustering and Deep Learning.

De novo peptide sequencing is a valuable technique in mass-spectrometry-based proteomics, as it deduces peptide sequences directly from tandem mass spectra without relying on sequence databases. This database-independent method, however, relies solely on imperfect scoring functions that often lead to erroneous peptide identifications. To boost correct identification, we present NovoRank, a postprocessing tool that employs spectral clustering and machine learning to assign more plausible peptide sequences to spectra. Prior to de novo peptide sequencing, spectral clustering is applied to group similar spectra under the assumption that they originated from the same peptide species. NovoRank then employs a deep learning model, incorporating both cluster-derived proteomic features and individual spectrum characteristics, to rerank the candidate peptides produced by de novo peptide sequencing. Our results show that NovoRank significantly enhances the performance of various de novo peptide sequencing tools, increasing both recall and precision by 0.020 to 0.080 at the peptide-spectrum match (PSM) level. Notably, NovoRank achieves a recall as high as 0.830 for Casanovo at the PSM level. The source code of NovoRank is freely available at https://github.com/HanyangBISLab/NovoRank and is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International.

Assessing handwriting skills in a web browser: Development and validation of an automated online test in Japanese Kanji

AbstractOnline language and literacy assessments have become prevalent in research and practice across settings. However, a notable exception is the assessment of handwriting and spelling, which has traditionally been conducted in person with paper and pencil. In light of this, we developed an automated, browser-based handwriting test application (Online Assessment of Handwriting and Spelling:OAHaS) for Japanese Kanji (Study 1) and examined its psychometric properties (Study 2). The automated scoring function using convolutional neural network (CNN) models achieved high recall (98.7%) and specificity (84.4%), as well as high agreement with manual scoring (95.4%). Additionally, behavioral validation with data from primary school children (N = 261, 49.0% female, age range = 6–12 years) indicated the high reliability and validity of our online test application, with a strong correlation between children’s scores on the online and paper-based tests (r = .86). Moreover, our analysis indicated the potential utility of writing fluency measures (latency and duration) that are automatically recorded by OAHaS. Taken together, our browser-based application demonstrated the feasibility and viability of remote and automated assessment of handwriting skills, providing a streamlined approach to research and practice on handwriting. The source code of the application and supporting materials are available on Open Science Framework (https://osf.io/gver2/).

Clone detection to prevent software piracy in android play store

Android is currently one of the most popular smartphone operating systems. Many application developers have been enticed by the enormous demand for mobile smartphone devices. The availability of reverse engineering tools for Android applications, however, also attracted virus authors' and plagiarists' attention. Cloning of applications has been a significant challenge to the Android market in recent years. However, Android accounts for the lion's share of all mobile malware globally, and its security vulnerabilities have garnered a lot of public attention. In this research, we look into how to recognize known Android malware using a clone detector. We compile a set of Android programs known to be malicious as well as a set of good programs. NiCad, a near-miss clone detector, is used to locate the classes of clones in a small fraction of the malicious programs after we retrieve the Java source code from the dex (Dalvik Executable file) file of the applications. The remaining malicious programs' source files are then searched for using these clone classes as a signature. As a control group, the non-harmful collection is utilized. In our analysis, we were able to decompile over 100 potentially harmful programs from 19 different malware families. According to our findings, a small sample of malicious programs can be used as a training set to detect 95% of known malware with a 96.88% accuracy rate and extremely few false positives. Our technique can successfully and consistently identify detrimental programs that are part of specific malware families. Moreover, illegal distribution is another part of software piracy, which is also a prevalent issue in the IT industry. We described an algorithm for the Android library licensing improvement to give insights to a developer on enhancing security measures to prevent the illegal distribution of individual Android applications.

Approximating spatial distance through Confront networks: application to the segmentation of medieval Avignon

Abstract In historical studies, the older the sources, the more common it is to have access to data that are only partial, and/or unreliable or imprecise. This can make it difficult, or even impossible, to perform certain tasks of interest, such as the segmentation of some urban space based on the location of its constituting elements. Indeed, traditional approaches to tackle this specific task require knowing the position of all these elements before clustering them. Yet, alternative information is sometimes available, which can be leveraged to address this challenge. For instance, in the Middle Ages, land registries typically do not provide exact addresses, but rather locate spatial objects relative to each other, e.g. x being to the North of y. Spatial graphs are particularly adapted to model such spatial relationships, called confronts, which is why we propose their use over standard tabular databases. However, historical data are rich and allow extracting confront networks in many ways, making the process non-trivial. In this article, we propose several extraction methods and compare them to identify the most appropriate. We postulate that the best candidate must constitute an optimal trade-off between covering as much of the original data as possible, and providing the best graph-based approximation of spatial distance. Leveraging a dataset that describes Avignon during its papal period, we show empirically that the best results require ignoring some of the information present in the original historical sources, and that including additional information from secondary sources significantly improves the confront network. We illustrate the relevance of our method by partitioning the best graph that we extracted, and discussing its community structure in terms of urban space organization, from a historical perspective. Our data and source code are both publicly available online.

DANI: fast diffusion aware network inference with preserving topological structure property

Numerous algorithms have been proposed to infer the underlying structure of the social networks via observed information propagation. The previously proposed algorithms concentrate on inferring accurate links and neglect preserving the essential topological properties of the underlying social networks. In this paper, we propose a novel method called DANI to infer the underlying network while preserving its structural properties. DANI is constructed using the Markov transition matrix, which is derived from the analysis of time series cascades and the observation of node-node similarity in cascade behavior from a structural perspective. The presented method has linear time complexity. This means that it increases with the number of nodes, cascades, and the square of the average length of cascades. Moreover, its distributed version in the MapReduce framework is scalable. We applied the proposed approach to both real and synthetic networks. The experimental results indicated DANI exhibits higher accuracy and lower run time compared to well-known network inference methods. Furthermore, DANI preserves essential structural properties such as modular structure, degree distribution, connected components, density, and clustering coefficients. Our source code is available on GitHub (https://github.com/AryanAhadinia/DANI).

REQUIREMENTS ENGINEERING CHALLENGES IN OPEN-SOURCE SOFTWARE DEVELOPMENT: A MULTIVOCAL LITERATURE REVIEW PROTOCOL WITH PRELIMINARY RESULTS

In the dynamic landscape of software development, open source software (OSS) has acquired traction due to its collaborative and transparent nature. In OSS, source code of the software is accessible to users under a copyright license. The license allows the users to examine, improve, alter and distribute the software either in modified or unmodified form. Richard Stallman, programmer at MIT, presented the idea of making source code freely accessible in the year 1983, from an ideological movement. Users must accept the conditions of a license before the use of OSS. Requirement engineering (RE) is the process of gathering requirements statements, along with the documentation and analysis for a project. Requirements that are insufficient, unclear, inconsistent, and incomplete, may adversely affect software development. In OSS, requirements emerge through online conversations, as well as discussions evolve continually in different online forums and emails. Requirements’ gathering is a complex process and its practitioners may face different challenges while practicing it. A Multivocal literature review (MLR) will be conducted to find out challenges, success factors, and practices/solutions for the identified challenges in RE process of OSS. This research will help academic researcher, OSS practitioners and managers to improve RE process, leading to more efficient, effective and successful software applications.