Varsha Bioscience and Technology (India)

This study presents the development and validation of an open-source finite element framework implemented in Python for the analysis of laminated composite plates using First Order Shear Deformation Theory (FSDT). The formulation employs graded quadrilateral meshing with eight-noded serendipity elements and incorporates full laminate constitutive modeling through computation of the A, B, and D stiffness matrices along with transverse shear effects. The framework enables automated geometry generation for plates with geometric discontinuities, global stiffness assembly, displacement solution, plylevel stress recovery, and contour visualization. To assess numerical accuracy, the developed solver is benchmarked against the layered shell formulation (SHELL181) in ANSYS under identical geometry, mesh density, material properties, boundary conditions, and loading. Comparisons are performed for in-plane and transverse displacements as well as normal stress components in laminated plies for a composite plate containing a central circular cutout. The Python-based results demonstrate strong agreement with the commercial solver, with displacement deviations below 1% and stress deviations within 5%. The study highlights the feasibility of using open-source computational tools for small- to medium-scale composite structural analyses, reducing dependency on high-cost licensed software for preliminary design studies and parametric investigations. Furthermore, the developed framework is extended toward the generation of high-fidelity finite element datasets to enable data-driven modeling. Specifically, the solver infrastructure is being utilized to create structured simulation databases for training Graph Neural Network (GNN)-based predictive models aimed at replacing computationally expensive finite element simulations. This integration establishes a pathway toward rapid surrogate modeling of composite structures while preserving physics-informed accuracy

Abstract Designing functional peptides with specific structural and biochemical properties is critical for applications in protein engineering and therapeutic discovery. However, most peptide design approaches rely on evolutionary or local sequence optimization methods, which are limited when adapting to peptides’ shorter length, high conformational flexibility, and unique physicochemical constraints. While recent structure-based inverse folding models have shown success for proteins, these models often underperform on peptides because sequence recovery alone is not a reliable indicator of stability or foldability in short, flexible backbones. To address this challenge, we introduce InversePep, a generative diffusion model for structure-based peptide inverse folding. InversePep learns the conditional distribution of sequences that can adopt a given backbone conformation, enabling direct generation of peptides tailored to target structural geometries. The framework integrates a geometric graph neural network to encode 3D backbone features with a Transformer-based sequence refinement module that iteratively denoises candidate sequences during diffusion. Trained on a diverse set of peptide backbones sourced from Propedia and SATPdb, InversePep effectively captures structural and biochemical diversity across peptide families. In systematic evaluations on held-out peptide structures and the PepBDB benchmark, InversePep achieves a mean TM score of 0.38 and a median of 0.28, outperforming ProteinMPNN and ESM-IF1 in generating geometry-consistent peptide sequences. In-silico folding analyses confirm that sampled peptides reliably adopt the target conformations. These results highlight InversePep’s capability for designing structurally stable and sequence-diverse peptides, demonstrating its potential in antimicrobial peptide discovery, peptide therapeutics, and molecular probe development.

Abstract Small interfering RNA (siRNA) therapeutics have extraordinary potential for targeted gene silencing. They mediate post-transcriptional gene regulation by binding to complementary messenger RNA (mRNA) sequences and degrading them, thereby preventing the production of unwanted proteins. Recent machine learning and deep learning frameworks for predicting siRNA efficacy have only achieved moderate success as these models solely rely either on handcrafted features or on sequential relations and therefore cannot capture the full complexity of siRNA-mRNA interactions. In this context, we propose SiaRNA, which uses a Siamese Neural Network for feature-derived representations and a bidirectional cross-attention mechanism for sequence-level relationships. It uniquely identifies mRNAs and their corresponding siRNAs as paired entities, allowing unified and context-aware modeling. Unlike previous models, which discard 2-nucleotide (2-nt) overhangs at the 3’ end while using 21-nt efficacy labels, SiaRNA both trains and tests on 21-nt sequences to ensure biologically consistent predictions. Our model sets a new performance benchmark, outperforming previous state-of-the-art models. SiaRNA is trained on the HUVK dataset achieving an accuracy of 0.881, while its generalization has been confirmed by testing on the independent Simone dataset. These results prove SiaRNA’s potential as a reliable and biologically accurate framework to guide siRNA design and improve therapeutic outcomes.

Maintaining a balanced diet and controlling medical conditions like obesity and diabetes require accurate food calorie estimation. Conventional techniques frequently call for manual input, which can be error-prone and time-consuming. Automated systems for estimating food calories have been made possible by recent developments in artificial intelligence and machine learning. But these systems usually need a lot of data, which makes privacy issues arise. One potential remedy is federated learning, which allows for cooperative model training without the need for centralized data collection. This chapter examines the use of federated learning in food calorie estimation, emphasizing its benefits, methods, and guiding principles.

The incorporation of machine learning (ML) into networking marks a significant leap, transforming traditional network management and optimization. ML algorithms, capable of learning from extensive datasets, have significantly enhanced various aspects of network operations, including traffic prediction, resource allocation, and security measures. This chapter offers a comprehensive overview of ML's application in networking, detailing its evolution, methodologies, and key benefits. Supervised learning models like linear regression and neural networks are utilized for accurate traffic prediction, whereas reinforcement learning aids in dynamic resource allocation and optimization. Additionally, anomaly detection and intrusion detection systems benefit from unsupervised learning techniques and advanced ML models, like decision trees and support vector machines. By exploring these ML techniques, this chapter highlights their critical role in improving network performance, security, and efficiency, demonstrating the profound impact of ML on modern networking.

Wireless Sensor Networks (WSNs) consist of spatially distributed sensor nodes that monitor and transmit data to the base station. Wireless sensor networks are widely used in environmental monitoring, smart cities, healthcare, and disaster management applications. However, the major disadvantage of WSNs is their excessive energy consumption, which shortens their lifetime and limits their data integrity. Traditional methods face redundant retransmissions and collision problems which lead to congestion and fast energy consumption. To overcome these problems, a novel Secure Clustering and Sleep-Wakeup based Energy Efficient Routing using Fennec fox optimized deep learning (SCS-EEF) framework has been proposed to reduce energy usage and increase network lifetime in WSN. Fuzzy C Means based Balanced Iterative Reducing and Clustering Using Hierarchies (Fuzzy-BIRCH) clustering is employed in the proposed framework to improve cluster formation and reduce communication costs. To enhance energy efficiency, Fennec Fox Optimization (FFO) is used for optimal cluster head selection (CHS) with fair energy distribution across nodes and to reduce premature node failures. An energy-saving dynamic sleep-wakeup schedule is proposed to eliminate redundant transmissions. The hybrid Temporal Convolutional Network-based Bidirectional Gated Recurrent Unit (TCN-BiGRU) network predicts multipath routes through the classification of data into emergency and non-emergency. The proposed framework reduces energy consumption by 39.3%, 29.2%, and 26.1% over HBWCO, IBORSDFFNL, and EER-CGHHOA, respectively. Furthermore, the proposed model reduces latency by 10.7%, 9.42%, and 7.4% over HBWCO, IBORSDFFNL, and EER-CGHHOA, respectively.

The Parking System is a digital platform designed to make car service centers work more smoothly and to give customers a better experience. Many service centers still rely on manual processes, which often lead to problems such as long booking times, poor scheduling, confusion in managing spare parts, and missing service records. This system solves these issues by providing useful tools for both customers and managers. For customers, it offers a simple online portal where they can quickly book appointments, check the full history of their vehicle’s services, and make secure online payments. This saves time, reduces hassle, and helps them keep track of their vehicle’s maintenance. For managers and staff, the Parking System provides an easy-to-use admin panel that brings all important tasks into one place. With this panel, managers can plan staff schedules, keep track of available services, monitor inventory, and manage customer bookings in an organized way. By reducing manual work, the system makes day-to-day operations faster, more accurate, and more efficient. The platform is built using modern technologies to ensure it is fast, reliable, and safe to use. The frontend is created with Next.js, giving users a smooth and responsive experience, while the backend is built with NestJS (Node.js) to handle tasks quickly and flexibly. Information is stored securely in PostgreSQL using Prisma, and trusted payment gateways are included to keep financial transactions safe. Overall, the Parking System improves customer satisfaction and helps service centers run more efficiently by saving time, reducing errors, and simplifying the management of services. It represents a step toward digital transformation in the automotive service industry, offering a smarter and more convenient way for people to access and manage vehicle services.

Predicting crop yield is a crucial factor in precision agriculture, especially under the conditions of increasing environmental uncertainty and climatic variability. This interest in predicting crop yield increased the necessity to develop advanced machine learning and dl models to refine the capabilities of accurate and efficient crop yield prediction. Existing models fail to handle real-time data and multiple types of data like weather conditions and IoT sensor readings. So it is crucial for us to implement a better crop yield prediction framework that can work well in all scenarios. In this paper, an Integrated Multi Stage Transformer (IMST) framework is discussed that uses an advanced Deep Learning (DL) approach. This IMST model uses a self-attention mechanism that captures complex spatial and temporal dependencies. The dataset contains different data like weather conditions, IoT sensor readings. The model is implemented and their performance is assessed using several evaluation metrics like accuracy, R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and F1 score. The transformer model results with an accuracy of 94.3%. It addresses the difficulties commonly found in real-world applications and exhibits good performance with real-time data.