Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{208491,
        author = {Aanisha S and Arjun N and Priyadharshini U and Mohamed Jamal Basha K and Dr P Kola Sujatha},
        title = {Unified Structure-Aware Sequence Framework for Protein-Peptide Interaction},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {4},
        pages = {1892-1900},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=208491},
        abstract = {Protein–peptide interactions play a fundamental role in numerous biological processes including cell signaling, immune response, molecular recognition, and drug targeting. Accurate identification of peptide-binding regions in proteins is essential for understanding biological mechanisms and designing therapeutic molecules. Existing computational approaches primarily focus on interaction prediction using sequence-based learning techniques, while providing limited structural interpretation and lacking mutation-aware analysis. This paper proposes a unified structure-aware sequence framework for protein–peptide interaction analysis using geometric deep learning and structural topology modeling. The framework integrates Evolutionary Scale Modeling (ESM-2) embeddings with Graph Neural Networks (GNNs) to capture both evolutionary and spatial characteristics of protein structures. A structure-aware groove localization module is introduced to identify potential peptide-binding grooves using residue-level geometric representations. To improve structural understanding, a Concavity-Gated Equivariant Graph Neural Network (CG-EGNN) is proposed, incorporating surface concavity and orientation information for geometry-aware message passing.
In addition, a mutation sensitivity analysis framework is developed to evaluate the effect of amino acid substitutions on groove probability, concavity, and residue orientation. A pocket classification module using DBSCAN clustering and an improved 1D CNN is employed to identify exact binding pocket regions from protein surface structures. Finally, an LLM-based explanation module using Mistral-7B-Instruct generates human-readable interpretations for mutation-induced structural changes.
Experimental evaluation on PepBDB and SKEMPI datasets demonstrates that the proposed framework effectively localizes peptide-binding grooves, predicts pocket regions with high accuracy, and provides meaningful structural interpretation of mutation effects.},
        keywords = {},
        month = {September},
        }