**5. Challenges and opportunities**

We have presented three important applications of MD simulation in drug design against influenza A virus. While MD simulation has been proven as a powerful tool complement to experiment in drug discovery processes, the problems in accuracy of drug parameter and limitation in time scale of MD simulation definitely affect applications of this tool. The challenge in drug force field development is that, unlike uniformed amino acids or nucleotides, each drug has its own structure and requires separate parameter. Since force field for general macromolecules have been well developed, attention should be shifted to development of adequately accurate drug force field (Wang et al., 2004; Zoete et al., 2012). Another challenge in MD simulation is limitation of time scale. Several important phenomena which are important for drug design application happen in the time scales that can't be done by even the biggest supercomputers in the world. Over the past decade several methods were developed to speed up classical MD simulation including SMD, accelerated MD (Hamelberg et al., 2009), coarse-grained MD simulation (Izvekov et al., 2005). Last but not least, the implementations of specialized hardware including CBE, GPU, and FPGA architectures are important to broaden application of MD simulation to modern drug discovery (Horacio et al., 2012). The recent finding of oseltamivir binding pathway using GPU acceleration and SMD (Le et al., 2010) is an example that the combined improvement of theoretical algorithm and specialized hardware architecture has enabled us to increase the success rate and decrease the cost in rational drug discovery.
