Research

Protein-Protein Interactions in Systems Biology

Most proteins function in coordinated teams through various pathways. Thus, protein-protein interactions (PPIs) are crucial to understand cellular functions, signaling pathways, and the mechanisms of diseases. Recent advances in deep learning and AI have shown remarkable potential in predicting PPI structures with accuracy. However, a major challenge lies in managing the immense computational demands of genome-wide predictions. Achieving a balance between computational efficiency and predictive accuracy remains a critical research area.

Building on our previous work with PrePPI-AF and ZEPPI, we are leveraging state-of-the-art deep learning models alongside traditional sequence analyses to re-evaluate the 30 billion protein-protein interaction models generated by PrePPI-AF. The developed methodology will be applied on predicting the Host-Pathogen Interactome. By unraveling the molecular mechanisms behind infectious diseases, we aim to advance the development of targeted therapies, vaccines, and strategies to combat both existing and emerging pathogens. Additionally, by combining sequencing data and omics-data analysis, we aim to gain deeper insights into the biological mechanisms that drive disease processes.

Protein Dynamics and Protein Design

Given the millions of predicted PPIs, we are interested in exploring the complex networks formed by experimentally known proteins. The objective is to identify novel pathways and achieve a complete mapping of these protein functions. Once specific target PPIs are identified, further molecular investigation into their binding dynamics and energetics will be conducted, with the final goal of designing optimized protein binders and small molecule inhibitors.

Protein Binding Affinity and Their Changes

Predicting protein binding free energy in silico is challenging, largely due to the relatively shallow energy minima associated with binding. In this context, we aim to develop a physics/chemistry enhanced ML/AI approach to quantitatively assess the binding free energy of proteins. Further, we seek to extend this model to predict how mutations, small molecules, or post-translational modifications influence binding free energy—areas often referred to as the 'dark matter' of biology.