Advancing Protein and Peptide Research through Multiscale Modelling and Machine Learning

Advancing protein and peptide research through multiscale modeling and machine learning offers transformative potential for understanding these vital biomolecules. Proteins and peptides play essential roles in biological systems, governing processes from cellular signaling to immune responses. Multiscale modeling enables researchers to capture protein and peptide behavior across different levels, from atomic interactions to larger biological assemblies, providing insights into structure, dynamics, and function that isolated methods cannot achieve. Concurrently, machine learning techniques leverage large datasets to predict protein-protein/peptide interactions, binding affinities, and functional sites with unprecedented accuracy. Together, but approached separately, these technologies empower detailed investigations, accelerate drug discovery, and aid in the design of novel biomolecules.

This Research Topic aims to tackle the challenge of accurately understanding and predicting the complex behaviors and interactions of proteins and peptides, which are critical for biological functions and therapeutic development. Despite advances in experimental techniques, capturing the full range of protein and peptide dynamics across different scales from atomic to cellular levels remains difficult due to computational limitations and the inherent complexity of biological systems. Furthermore, interpreting vast biological data to predict interaction networks and functional sites demands sophisticated computational tools. To address these challenges, the Research Topic will focus on leveraging multiscale modeling to reveal detailed mechanistic insights by simulating molecular processes at appropriate resolutions. Additionally, it will also highlight the power of machine learning to analyze large-scale biological data, predict binding affinities, and identify functional motifs with high accuracy. Recent advances, such as graph neural networks, deep learning frameworks for peptide-protein interactions, and enhanced sampling techniques in multiscale simulations, offer promising avenues to overcome these barriers. This combined focus can enable more precise, interpretable, and scalable models that accelerate discoveries in protein and peptide science.

Themes of interested include, but are not limited to:

- Investigating protein and peptide structure and dynamics through multiscale modeling at atomic to mesoscale levels.

- Developing machine learning algorithms for predicting peptide-protein interaction sites and binding affinities.

- Exploring graph neural networks and deep learning techniques specific to biomolecular sequence and structural data.

- Applying coarse-grained and enhanced sampling methods in multiscale simulations to improve computational efficiency.

- Integrating computational predictions with experimental validation for biomolecular interactions.

- Addressing interpretability and explainability of machine learning models in the context of protein and peptide function and design.

- Use of ML and AI algorithms for the development of empirical (low resolution or atomistic) force fields for biomolecules