Identification of novel key genes and the molecular network underlying a pathological or physiological process is essential but still very changing for biomedical communities. I will present novel strategies to infer reliable molecular networks from high-time-resolution dynamic data. Based on these reliable networks, we predict novel key genes. In this work, I will demonstrate several examples showing the high predictive power of our network-based strategies in regulatory T cells, HIV-resistant T cells as well as several pathogenic strains. I will try to convince you that our network strategies can be generally applicable in not only a better understanding of molecular mechanisms, but also novel drug-target identification.