Software vulnerabilities pose significant risks to the security and reliability of modern systems, making automated vulnerability detection an essential research area. Traditional static and rule-based approaches are limited in scalability and adaptability, motivating the adoption of data-driven methods. In this survey, we present a comprehensive review of Machine Learning (ML), Deep Learning (DL), and Large Language Models (LLMs) techniques for vulnerability detection. We analyze recent advances in feature representation, fine-tuning strategies, generative approaches, and prompt engineering, while highlighting their ability to capture both syntactic and semantic properties of source code. Furthermore, we examine commonly used evaluation metrics and provide a critical discussion of key challenges, including the lack of large-scale real-world datasets, limited vulnerability coverage, class imbalance, interpretability gaps, hallucination, and high computational costs. To address these issues, we outline promising future research directions, such as neuro-symbolic hybrid methods, parameter-efficient fine-tuning, continual learning, cross-language generalization, and explainable AI for vulnerability detection. Unlike previous studies, the present work explores learning paradigms from ML to LLMs using comprehensive evaluation criteria that highlight analytical capability, feature interpretability, and code-context comprehension. By combining these factors, our study addresses the methodological gap between classic feature-based approaches and current LLM-driven reasoning frameworks, providing beneficial insights to develop robust, scalable, and trustworthy software vulnerability detection systems.

Antimicrobial peptides (AMPs) play a vital role in the immune defence systems of various organisms and have garnered significant attention for their potential applications in biotechnology and medicine. There are several approaches to identifying AMPs including clinical isolation and characterization, functional genomics, microbiology techniques, and others. However, these methods are mostly expensive, time-consuming, and require well-equipped labs. To overcome these challenges, machine learning models are a potential solution due to their robustness and high predictive capability with less time and cost. In this study, we explored the efficacy of stacking-based ensemble machine-learning techniques to identify AMPs with higher accuracy and precision. Five distinct feature extraction methods, namely Amino Acid Composition, Dipeptide Composition, Moran Autocorrelation, Geary Autocorrelation, and Pseudo Amino Acid Composition were employed to represent the sequence characteristics of peptides. To build robust predictive models, different traditional machine learning algorithms were applied. Additionally, we developed a novel stacking classifier, aptly named StackAMP, to harness the collective power of these algorithms. Our results demonstrated the exceptional performance of the proposed StackAMP ensemble method in antimicrobial peptide identification, achieving an accuracy of 99.97%, 99.93% specificity, and 100% sensitivity. This high accuracy underscores the effectiveness of our approach, which has promising outcomes for the rapid and accurate identification of antimicrobial peptides in various biological contexts. This study not only contributes to the growing body of knowledge in the field of antimicrobial peptide recognition but also offers a practical tool with potential applications in drug discovery, biotechnology, and disease prevention.

Infectious fungi have been an increasing global concern in the present era. A promising approach to tackle this pressing concern involves utilizing Antifungal peptides (AFP) to develop an antifungal drug that can selectively eliminate fungal pathogens from a host with minimal toxicity to the host. Accordingly, identifying precise therapeutic antifungal peptides is crucial for developing effective drugs and treatments. This study proposed MLAFP-XN, a neural network-based strategy for accurately detecting active AFP in sequencing data to achieve this objective. In this work, eight feature extraction techniques and the XGB feature selection strategy are utilized together to present an enhanced methodology. A total of 24 classification models were evaluated, and the most effective four have been selected. Each of these models demonstrated superior accuracy on independent test sets, with respective scores of 97.93 %, 99.47 %, and 99.48 %. Our model outperforms current state of the art methods. In addition, we created a companion website to demonstrate our AFP recognition process and use SHAP to identify the most influential properties.

Bitter peptides are short amino acid chains that produce a bitter taste. These peptides are made primarily in food processing through the chemical reduction of peptides. The bitterness arises from the specific sequence of amino acids in peptides, which interact with the bitter taste receptors on the human tongue. These peptides influence nutrition and health, offering insights into protein digestion and bioactive advantages. Hence, correctly identifying bitter peptides is pivotal for revealing the biochemical properties of efficient medication. The computational approach is most suitable for identifying bitterness, where most studies obtained insufficient outcomes. Therefore, the current study developed an ensemble-based framework called "BitterEN", where we integrate the Gradient Boosting (GB) and Multi-layer Perception (MLP) methods. Our proposed method improved more than 3 % of accuracy compare to all of the state-of-the-arts methods, where the proposed approach achieved 0.995 accuracy in merged feature extractions with the Random Forest (RF) feature selection method. We used 50 iterations over the performance evaluation phases to enable a more exact generalization of model performance. In addition, we provided a convenient GitHub-based version of our bitter peptide identification. It highlights the practical applicability of these findings. We are optimistic that the proposed approach might benefit many fields, including healthcare development and nutritional science.

Laboratory-based detection of D sites is laborious and expensive. In this study, we developed effective machine learning models employing efficient feature encoding methods to identify D sites. Initially, we explored various state-of-the-art feature encoding approaches and 30 machine learning techniques for each and selected the top eight models based on their independent testing and cross-validation outcomes. Finally, we developed DHUpredET using the extra tree classifier methods for predicting DHU sites. The DHUpredET model demonstrated balanced performance across all evaluation criteria, outperforming state-of-the-art models by 8 % and 14 % in terms of accuracy and sensitivity, respectively, on an independent test set. Further analysis revealed that the model achieved higher accuracy with position-specific two nucleotide (PS2) features, leading us to conclude that PS2 features are the best suited for the DHUpredET model. Therefore, our proposed model emerges as the most favorite choice for predicting D sites. In addition, we conducted an in-depth analysis of local features and identified a particularly significant attribute with a feature score of 0.035 for PS2_299 attributes. This tool holds immense promise as an advantageous instrument for accelerating the discovery of D modification sites, which contributes too many targeting therapeutic and understanding RNA structure.