This project advances protein function prediction by integrating protein language models (PLMs) and graph convolutional networks (GCNs), addressing the limitations of traditional methods that rely heavily on sequence similarity. The proposed model leverages diverse protein features, including sequences, protein-protein interaction (PPI) networks, and InterPro domains, to create a robust computational framework. Utilizing the Evolutionary Scale Modeling (ESM-1b) PLM, high-dimensional feature embeddings are generated from protein sequences. These are integrated with PPI network data and InterPro domains through a two-layer GCN, enabling the model to capture complex interdependencies. The model’s performance was evaluated using metrics such as Fmax and Area Under the Precision-Recall Curve (AUPR) across different Gene Ontology (GO) categories: Molecular Function (MFO), Biological Process (BPO), and Cellular Component (CCO). The findings demonstrate that the model outperforms the most advanced techniques currently in use for BPO and CCO forecasts. However, MFO predictions require improvement, suggesting that future efforts should concentrate on more accurately identifying sequence-specific motifs. The study problem and objectives are presented first in the report, which is structured to give a thorough overview. A review of the literature, the research technique, and a detailed analysis of the experimental data are then included. The study concludes with reflections, highlighting areas for future research and the broader implications for biomedical and biotechnological applications.