Miniproteins, defined as polypeptides comprising fewer than 100 amino acids, represent a unique and increasingly important class of biomolecules with remarkable structural diversity and functional versatility. Despite their compact size, these molecules can adopt well-defined, stable three-dimensional structures, including α-helical, β-sheet, and hybrid folds. Such structural adaptability enables them to participate in a wide range of biological processes, including signalling pathways, regulation of transcription and translation, molecular recognition, and antimicrobial defence. Their small size, combined with high stability and adaptability, has positioned miniproteins as promising candidates in therapeutic design, protein engineering, and synthetic biology.o
A significant proportion of protein sequences identified across bacterial, archaeal, and eukaryotic genomes remains poorly characterised and is often annotated as hypothetical proteins. Within this largely unexplored sequence space lies a substantial number of potential miniproteins whose functions are yet to be determined. Recognising this gap, our research focuses on systematically identifying and characterising these hypothetical miniproteins using an integrated computational approach. The objective is not only to expand the current understanding of miniprotein biology but also to uncover novel candidates with potential functional and therapeutic relevance.
To achieve this, we are developing a comprehensive, multi-tiered computational pipeline. The process begins with large-scale genomic data mining to identify protein-coding sequences, followed by the extraction of unannotated entries that fall within the size range of miniproteins. These sequences are then subjected to antimicrobial peptide (AMP) prediction tools, enabling their classification into AMP-like and non-AMP-like groups. The AMP-like candidates are further analysed through homology modelling to generate structural models, which are subsequently refined and evaluated using molecular dynamics simulations. These simulations provide critical insights into conformational stability, residue-level flexibility, and potential interaction behaviour, offering a dynamic perspective that complements static structural predictions.
Building upon these analyses, we propose an artificial intelligence-driven framework that integrates sequence, structural, and dynamic features to automate the discovery and functional annotation of miniproteins. This approach is designed to enhance prediction accuracy, reduce manual intervention, and accelerate the identification of biologically relevant candidates. By leveraging advances in machine learning and computational biology, the framework aims to streamline the transition from raw genomic data to functionally characterised molecules.
Overall, this work establishes a scalable and systematic strategy for exploring the undercharacterised landscape of hypothetical miniproteins. By combining bioinformatics, structural modelling, molecular simulations, and AI-based analysis, our research contributes toward the rapid discovery of novel functional molecules, with particular emphasis on next-generation antimicrobial peptides. This integrated approach not only advances fundamental understanding but also holds significant promise for the development of innovative therapeutic solutions.