Seq. Length | Phylip | Paup | Naive | Fastdist |
|---|
10,000 | 5.47 s | 2.69 s | 0.24 s | 0.03 s |
100,000 | 60.2 s | 4 m25 s | 2.32 s | 0.27 s |
1,000,000 | - | 371 m50 s | 23.2 s | 2.71 s |
- Comparing the running time o Phylip s dnadist, Paup, our optimized Implementation of the naive approach, and the new algorithm fastdist. The test's are run on simulated data with 100 species. It should be noted that Phylip could not be run on sequences of length 1,000,000. Except for Paup which uses the analytic formula, the programs computed the K2P distance with a globally fixed transition transversion ratio of 2. Reconstructing a tree of 100 taxa using NJ takes no more than 0.01 seconds. Moreover, simply reading a file containing 100 sequences of length one million with the wc unix utility takes 0.75 seconds.
