StackPIP: An Effective Computational Framework for Accurate and Balanced Identification of Proinflammatory Peptides

Proinflammatory peptides (PIPs) play a crucial role in immune response modulation by orchestrating cytokine release and leukocyte recruitment. Accurate identification of PIPs is essential for understanding inflammation-related diseases and developing therapeutic interventions. Traditional experimental methods for PIP identification are labor-intensive and low-throughput, necessitating the development of robust computational approaches. In this study, we propose StackPIP, a novel machine learning framework that leverages a stacking-based ensemble strategy to enhance PIP prediction. StackPIP integrates multiple peptide descriptors capturing compositional, order, and physicochemical properties, coupled with 12 machine learning algorithms to construct a high-performing computational framework. Experimental results demonstrate that StackPIP outperforms existing computational methods, surpassing the accuracy of previous state-of-the-art approaches by nearly 5% while achieving balanced prediction results. Furthermore, an interpretability analysis was conducted to elucidate the critical sequence characteristics contributing to the proinflammatory activity. To facilitate accessibility, we have developed a user-friendly web server, enabling researchers to efficiently utilize StackPIP for PIP identification, which is freely available at https://awi.cuhk.edu.cn/~biosequence/StackPIP/index.php.