Protein Function Prediction Based on Protein-Protein Interactions: A Comparative Study

Since interactions among proteins are of central importance for virtually every process in living cells, their information improves our understanding of diseases and can provide the basis for new therapeutic approaches. The study of protein function prediction based on protein-protein interactions (PPI) is one the most important issues in the field of bioinformatics since it is crucial for the understanding of cell activities. In this paper, a comparative study of different methods for protein function prediction based on PPI is presented. Five selected methods including the neighbor counting, Chi-square, Markov random field, Prodistin, and weighted interactions methods are applied to yeast proteome and their prediction performance is compared. Results revealed variant differences in the sensitivity and specificity of prediction for the functional category: cellular role. The weighted interactions method showed the best sensitivity values, while the Chi- square method resulted in the worst sensitivity/specificity values.