Yıldız Technical University Research Information System
Not a member yet
91324 research outputs found
Sort by
A superhydrophobic magneto-flow system for semi-automated electrochemical detection of Salmonella Typhimurium in food samples utilizing an aptamer-based technique
No full textThe existing methodologies for detecting foodborne pathogens often involve laborious and time-consuming procedures, underscoring the need for rapid and effective detection techniques. Herein, we present a superhydrophobic magneto-flow system integrated with an electrochemical aptasensor designed for simple, sensitive, yet controlled detection of Salmonella enterica serovar Typhimurium (Salmonella) in food samples. A superhydrophobic surface created through a chemical gold reduction method on a polydimethylsiloxane substrate enabled a distinct fluid orientation within a microfluidic device that prevents merging, thereby reducing contamination and enhancing detection rates. A laminate magneto-driven microfluidic device was fabricated using laser cutting and assembled to establish channel barriers that allow for precise control of magnetic flow direction using magnetic force, rather than relying on external pumps or absorbent pads. Aptamers specifically targeting Salmonella were conjugated to magnetic beads and aligned on the superhydrophobic surface with a custom-designed pattern to facilitate one-time spotting of the requisite reagents. Using a single introduced platform, the entire operation, including Salmonella-aptamer capturing, washing, and electrochemical measurement, can be performed sequentially, akin to a move-pause station. The aptasensing system operated in a label-free format, demonstrated the capability to detect Salmonella in food samples, achieving high sensitivity successfully (limit of detection = 10.01 CFU/mL), selectivity, a rapid response time (30 min), and a broad linearity range (10 – 105 CFU/mL), verified by standard approaches. This device embodies a cutting-edge immunosensing platform that facilitates precise operational control and features simple fabrication processes, rendering it a viable alternative suitable for complex multi-step procedures
Comparing the Interactions of Trichomonas vaginalis/gallinae Legumain-Like Cysteine Protease 1 (LEGU-1) and Human Legumain (LGMN) Protein Sequences with Proton Pump Inhibitor Drugs (Lansoprazole, Omeprazole, and Esomeprazole) by Bioinformatics Analyses
No full textPurpose: The flagellar parasite Trichomonas vaginalis is the main cause of trichomoniasis cases globally and is associated with a broad range of complications. Due to the diverse range of virulence factors participating in the attachment, proliferation and resistance of this pathogen, preventive and well-tolerated compounds are necessary. One of the virulence factors in T. vaginalis, the legumain-like cysteine protease LEGU-1 is of particular interest as a target due to its potential influence on trichomoniasis and tumor development in urogenital systems, as well as its closely related to the avian strain T. gallinae. Previous studies on antineoplastic proton pump inhibitors revealed they also have legumain (LGMN) inhibitory activities. Methods: Therefore, this study aimed to compare the molecular interactions of T. vaginalis/gallinae LEGU-1 and H. sapiens LGMN with proton pump inhibitor drugs (lansoprazole, omeprazole, and esomeprazole) through sequence analysis, 3D modeling, and molecular docking. Results: Although sequence analyses revealed low homology between T. vaginalis/gallinae LEGU-1 and H. sapiens LGMN, secondary and 3D structural comparisons uncovered their structural conservation. Possible binding sites in all three proteins identified via CB-DOCK2 were compared to the previously described sites for LGMN, followed by targeted docking using Autodock Vina. Identification of amino acids mutually interacting with all three ligands by both programs revealed the overall conservation of the binding pockets. The variations in the number of amino acids within the binding sites for all three proteins displayed the variations in the binding energies for each ligand. Lansoprazole, omeprazole and esomeprazole were shown to bind T. vaginalis/gallinae LEGU-1 and H. sapiens LGMN, with lansoprazole having the highest binding energy. Conclusion: Conclusion Beyond our promising bioinformatics results, this study can guide further research on the development of alternative therapeutic methods against trichomoniasis and concomitant conditions
Gold-nanoparticle/copolymer-modified screen printed carbon based electrode nonenzymatic electrochemical sensor for sensitive detection of pyocyanin as a Pseudomonas aeruginosa infections biomarker
No full textPseudomonas aeruginosa causes severe infections such as wound, respiratory, and urinary tract infections. Pyocyanin, a redox-active molecule secreted by this bacterium, plays a key role in its pathogenicity and can serve as a biomarker for early infection diagnosis. In this study, we developed an electrochemical sensor exploiting the redox activity of pyocyanin for its sensitive detection. In this context, firstly, the surface printed electrode Poly(Aniline-co-Pyrrole)/Au-NP/SPCE electrode was obtained by coating gold nanoparticles onto the electrode produced by the electrocopolymerization of pyrrole and aniline together by the chronoamperometric method. A biosensor was designed to detect pyocyanin molecule in phosphate buffer solution (pH 7.4) solution using the prepared Poly(Aniline-co-Pyrrole)/Au-NP/SPCE electrode. The electrochemical performance of the hybrid electrode was evaluated using cyclic voltammetry, differential pulse voltammetry and electrochemical impedance spectroscopy analyses. The structural and morphological properties of the prepared electrode were investigated using Fourier transform infrared spectroscopy, scanning electron microscopy, X-ray diffraction and Raman analyses. A stable electrochemical biosensor was developed for rapid and sensitive detection of pyocyanin in phosphate buffer (pH 7.4). The sensor exhibited a linear response between 0.5 μM and 250 μM with a detection limit of 101.88 nM and a quantification limit of 339.59 nM. The repeatability test of the sensor electrode was performed, and the relative standard deviation value was found to be 5.14 %. A Bacterial culture containing the pyocyanin molecule was used to test the biosensor's applicability in a real sample. The results of the study demonstrate the potential use of electrochemical biosensors in the diagnosis of P. aeruginosa infections. Furthermore, this study is expected to shed light on future studies using clinical patient samples (blood, saliva, etc.)
Whisper, Translate, Speak, Sync: Video Translation for Multilingual Video Conferencing Using Generative AI
No full textThis paper addresses the growing need for seamless communication in multilingual video conferencing by presenting a novel, computationally efficient methodology for real-time video translation. While advancements in neural networks have enabled accurate speech translation and voice cloning, integrating these with lip synchronization for realistic talking head generation remains a challenge, particularly for real-time applications. This paper introduces a comprehensive video translation pipeline leveraging open-source deep learning models. We further propose a scalable system architecture incorporating a “Token Ring” mechanism to manage speaker turns and minimize computational load, addressing key challenges related to latency, scalability, and personalization in multilingual settings. A segmented batched processing protocol with inverse throughput thresholding and overlapping buffering is implemented to achieve near real-time performance. A simplified, universal prototype is developed to demonstrate the feasibility and efficacy of our approach, providing a foundation for building next-generation multilingual video conferencing systems. This work offers a practical framework for developers and businesses aiming to create inclusive and effective communication platforms
Development of Spraying Based Liquid Phase Microextraction Method for the Preconcentration of Flibanserin From Urine Samples via GC–MS Analyses
No full textFlibanserin is the initial pharmaceutical treatment for hypoactive sexual desire disorder (HSDD). The analysis of urine samples plays a crucial role in the quantitation of flibanserin since a portion of flibanserin is excreted unchanged in the urine. An analytical method was proposed to quantify flibanserin in artificial urine samples (as model matrices). The integration of the spray-assisted fine droplet formation-liquid phase microextraction (SFDF-LPME) method and gas chromatography–mass spectrometry (GC–MS) system was performed for the first time to improve the sensitivity of the GC–MS system for flibanserin. Several parameters, including spraying cycle, extraction solvent type, mixing type and period, and sample volume, were systematically optimized to enhance the signal-to-noise ratio (S/N) of the analyte. After determining the optimal conditions, the analytical performance measurements of the system were figured out. The limit of detection (LOD), the limit of quantification (LOQ), and coefficient of determination (R2) values were 6.91, 23.05 μg kg−1, and 0.9989, respectively. Recovery experiments were performed in artificial urine samples within the specified linear working range of 33.15–505.66 μg kg−1. The SFDF–LPME–GC–MS method was efficiently applied to artificial urine samples by computing the matrix-matching calibration strategy, with percentage recovery values ranging from 90.0% to 105.9%
Repository-Level Code Understanding by LLMs via Hierarchical Summarization: Improving Code Search and Bug Localization
No full textBug localization and semantic code search within large software repositories is a significant and time-consuming challenge for developers, particularly when dealing with bug reports from end-users who lack technical expertise. Traditional similarity-based code search methods struggle with the inherent domain and vocabulary mismatch between end-user reports and codebase semantics, while directly applying Large Language Models (LLMs) is hampered by their limited context windows and lack of repository-level understanding. To address these limitations, this paper introduces a novel, structure-aware methodology for creating repository-aware LLMs using hierarchical summarization. Our approach comprises a pre-processing phase that constructs an abstract repository tree, creates a context-aware LLM primed with project knowledge, and generates hierarchical summaries at project, directory, and file levels. The inference phase employs a top-down search strategy, guiding the LLM to progressively narrow down the search space from directory-level to file-level, effectively localizing bug-relevant code. This method mitigates the context window bottleneck and leverages LLMs’ semantic understanding to overcome domain gap issues. Evaluated on a real-world dataset of Jira issues from a large-scale industrial project, our approach significantly outperforms both Flat Retrieval baselines and state-of-the-art LLM + Retrieval-Augmented Generation (RAG) systems, achieving a Pass@10 of 0.89 and Recall@10 of 0.33. The results demonstrate the efficacy of hierarchical summarization in enabling scalable, task-agnostic, and structure-aware repository-level code comprehension for improved bug localization and code search, particularly in scenarios involving non-technical end-user bug reports
KENT KİMLİĞİNİ YANSITACAK VE İKLİM DEĞİŞİKLİĞİNE UYUM SAĞLAYABİLECEK ODUNSU BİTKİLER: ÇANAKKALE
No full textKENT KİMLİĞİNİ YANSITACAK VE İKLİM DEĞİŞİKLİĞİNE UYUM SAĞLAYABİLECEK ODUNSU BİTKİLER: DENİZLİ
No full textCombining 3D Urban Objects From All Around the World to Improve Object Classification and Semantic Segmentation
No full textGiven thegrowing number of applications in autonomous driving and urban digital twins,the development of effective solutions for urban point cloud classification isof extreme interest for the R&D community and commercial sector.State-of-the-art neural networks commonly lack adequate cross-datasetgeneralisation ability, mainly due to varying sensors and data collectionplatforms, object shape differences, as well as the presence ofunder-represented objects and imbalanced classes, especially with dense andhigh-resolution reality-based 3D data. This work demonstrates how the recentlyreleased ESTATE (A large dataset of under-represented urban objects) dataset(https://github.com/3DOM-FBK/ESTATE), full of thousands of under-representedurban objects such as traffic lights, electrical poles, pylons, and ventilationunits spread over 13 classes, can improve the performance of state-of-the-artpoint cloud classification algorithms. Experiments with different neuralnetworks and several testing configurations with sensor-specific inputs(coordinate, intensity, and colour) show the effectiveness of this dataset inenhancing the classification capabilities and increasing cross-datasetgeneralisation. Moreover, not only the adaptation of object classificationnetworks to the semantic segmentation pipeline is introduced, but also theimprovement of semantic segmentation performance by increasing the distributionof under-represented classes with the ESTATE dataset. Thanks to 3D urbanobjects from all around the world in the ESTATE dataset, the model’sapplicability for classifying an entirely different dataset is alsodemonstrated.</p