PLiCat: Decoding protein-lipid interactions by large language model

Abstract

MotivationProtein-lipid interactions are essential for many cellular processes, as proteins associate with diverse lipid molecules to exert distinct functions. However, existing approaches are limited in discriminating among lipid categories in these interactions. Recent advances in protein language models enable discovery of novel sequence insights. ResultsWe introduce PLiCat (Protein-Lipid interaction Categorization tool), a predictive framework designed to classify lipid categories interacting with proteins. By integrating ESMC and BERT models into a hybrid architecture, PLiCat achieves accurate and interpretable predictions to distinguish lipid-binding signatures across eight major lipid categories. Furthermore, we investigated the potential of PLiCat to identify lipid-binding sites and assess the impact of pathogenic mutations on lipid-binding events. Collectively, PLiCat provides a powerful framework for elucidating the lipid-binding codes, offering new opportunities for exploring lipid-binding specificity and guiding rational protein design. Availability and implementationThe PLiCat source code and processed datasets are available at https://github.com/Noora68/PLiCat.