Biological mechanisms consist of protein interaction networks. However, most of protein interaction prediction are based on interspecies biological evidences (knowledge-based predictions) or exhibits low accuracy for weak interactions and requires high computational power (in silico docking method). In this study, I suggest a novel method to identify protein interaction using interaction energy distribution. Protein interactions with specific partners can be predicted by searching narrow funnel-like interaction energy distribution. Kinases and E3 ubiquitin ligase are important signaling components. However, as their interactions are weak and various, those interaction predictions are limited. I described that their interaction partners can be identified by analyzing interaction energy distribution with low computational cost. The accuracy of this prediction was similar or even higher than those of yeast two-hybrid screening, the most widely applied interaction screening method. Furthermore, I showed that other specific protein interactions have narrow funnel-like interaction energy distribution. In summary, this knowledge-free protein interaction prediction method would broaden our understanding of protein interaction networks.