Defence Metallurgical Research Laboratory

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.

Thermomechanical Metamaterials The cold spray technique, primarily meant for component repair applications, preserves the original material properties owing to non-melting, producing dense, well-adhered coatings with minimal oxidation or thermal distortion. The image illustrates the powder metals being deposited on the substrate from a spray gun, with a magnified inset showing the jetting behavior of individual particle during high-energy impact. The orientation maps of the powders in-flight and after coating, highlights the microstructural evolution in each impacted particle viz-a viz the whole coating. More information can be found in the Research Article by Malar Vadani, Sabeur Msolli, Ayan Bhowmik, and co-workers (10.1002/adem.202501848).

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.

This work examines the effects of prolonged heat‐treatment on the microstructure and mechanical characteristics of cold spray (CS) IN718. The heat treatment reduced coating porosity from 3% to 0.8%. The ageing treatment results in a synergistic effect of precipitation hardening from intermetallic phases and carbides, with a simultaneous reduction in the work‐hardened microstructure through recovery, manifested as reduced dislocation density in the coating. The as‐sprayed coatings exhibit reduced cohesive strength and toughness, along with a significant anisotropy in tensile characteristics, which is a primary objective of the study. The causes of the mechanical anisotropy in the CS coatings have been elucidated. The postprocessing resulted in an increase of ≈120% in cohesive strength and ≈75% in ductility of the coating due to the aforementioned alterations in microstructures. The tensile response of the coatings in both as‐sprayed and heat‐treated conditions has been analysed using numerical modelling, which utilizes crack widths and their densities within the coating. The model clearly shows that cracks perpendicular to the loading affect tensile strength. The model agrees with experimental tensile strength data and accurately predicts crack widths and densities in each anisotropic direction. The finite element models reveal an empirical connection between tensile strength and crack size.

The rapid growth of digital documents has created a pressing demand for intelligent retrieval systems that move beyond keyword-based search and provide contextually accurate information. Existing methods often fail to capture semantic relationships, leading to incomplete or irrelevant results. To overcome these limitations, this paper proposes an interactive PDF chatbot that enables natural language querying through a conversational interface. The framework integrates three key components: (i) semantic embeddings generated using Sentence-BERT to capture contextual meaning, (ii) scalable similarity search with FAISS indexing for efficient retrieval, and (iii) context-aware response generation using the FLAN-T5- XL model. The novelty of this work lies in the integration of semantic understanding with efficient vector-based retrieval and generative modeling, offering improved accuracy and contextual relevance compared to conventional keyword-driven approaches. The system is implemented as a web application that allows users to upload documents and obtain precise, coherent responses without requiring technical expertise. Experimental evaluation demonstrates that the proposed model achieves higher retrieval accuracy and generates more contextually appropriate answers compared to baseline approaches, thereby highlighting its potential utility for researchers, professionals, and other users engaged in document-intensive tasks.

A modified two-step sintering process is compared with the conventional two-step sintering process for sintering alumina. In modified two-step sintering, the first step was heated with microwave-assisted conventional heating up to 1500 °C ( T 1 ) and the second step was heated at 1450 °C ( T 2 ) with conventional heating alone. T 1 and T 2 were maintained the same for both the modified two-step sintering and conventional two-step sintering processes. Results showed that modified two-step sintering produced marginally lesser grain growth but similar densification. A near-full densification (>99%) was obtained on the holding sample for 3 h at T 2 in both the two-step sintering processes. A minimum grain size of 1.7 and 1.9 µm and a maximum grain size of 3.0 and 3.2 µm were obtained at 0.5 and 3 h sintering durations, respectively, for both processes. Modified two-step sintering produced a Vickers hardness of 18.7 ± 0.2 GPa, and the conventional two-step sintering process produced a hardness of 18.5 ± 0.2 GPa.

This study investigates blast wave interactions with different geometries to support safer and more effective design strategies for blast mitigation. Controlled shock tube experiments (driver length: 0.8 m, driven length: 6 m, and helium driver gas) were used to replicate blast wave conditions without live explosives. Two-dimensional simulations with an in-house multi-component Navier–Stokes solver were performed for rectangular, circular, and double-wedge objects placed at x/D=2.5 and 5 under pressure ratios of 13 and 57. Flow features were analyzed using numerical Schlieren, vorticity, pressure, and density fields, along with object load histories. The results show dominant vorticity generation over rectangular objects, while blast wave reattachment is governed primarily by standoff distance rather than strength-shortest for the double wedge, followed by circular and rectangular objects. These findings provide valuable benchmarks for validating computational models and predicting real-world blast effects.

A steel pin used as a locator for internal nose cap of long range cruise missile was found in cracked condition after it was subjected to transportation, vibration and bump tests. Metallurgical investigation was carried out with the failed pin. The crack was found to be inter-granular and a thin black oxide layer was observed on the fracture surface. Based on the evidence, it is concluded that the pin failed because of quench cracking and the crack was initiated prior to tempering and zinc plating.