We often encounter huge data in sequence analyses. With rapidly increasing sequence data, this tendency will be stronger in future than it is today. Management of such huge data will be a critical problem. Firstly, parallel computing provides a direct solution to this problem. Secondly, in reliable but complicated data analysis methods such as the maximum likelihood method, parallel computing has the advantages to reduce computational time.
On the other hand, parallel computing has the following problems. Parallel computers are not easily available. Even if a parallel computer is available, they may be used by too many users to obtain efficient performance. Distributed computing such as PVM [Geist et al., 1998] is the alternative solution for parallel computing, however, communication cost may be critical in this case. In addition, we should note that today's parallel computers are a little bit unstable because of technical problems.
Generally speaking, parallel programming is complicated in procedural language like C. On the contrary, we are free from such difficulty in programming with DeepForest.