Tree Model T1 from Strimmer and von Haeseler. a and b are rational numbers such that a < b.
We generate 500 replications with the Jukes-Cantor model via a software evolver from PAML package. How often a tree reconstructed via MJOIN has the same tree topology with the true tree?
The number represents a percentage which we got the same tree topology.
The numbers in the table mean succes rate for the tree model T1 for each a and b.
(l is the length of datasets and m is the number of leaves for each subtree).
Note: All CPU time for MJOIN with each a/b is less than 3 seconds.
l | a/b | m=2 | m=3 | m=4 | fastDNAml |
500 | 0.01/0.07 | 68.2 | 76.8 | 80.4 | 74.8 |
0.02/0.19 | 54.2 | 61.2 | 73.6 | 55.6 | |
0.03/0.42 | 10.4 | 12.6 | 23.8 | 12.6 | |
1000 | 0.01/0.07 | 94.2 | 96 | 97.4 | 96.6 |
0.02/0.19 | 87.6 | 88.6 | 96.2 | 88 | |
0.03/0.42 | 33.4 | 35 | 52.4 | 33.6 |