Skip to main content
. 2013 Mar 15;8(3):e58537. doi: 10.1371/journal.pone.0058537

Table 3. A comparison of the performance of different tree inference methods following trimming of realigned simulated sequences.

GBLOCKS trimmed FFT-NS-1 RFd FP FN Precision Recall
PhyML 4G SPR (SH 50) (3) 68.4 a 21.8 a 46.5 a 0.896 a 0.813 a
RAxML (3) 74.1 a 37.0 c 37.0 a 0.851 b 0.851 a
PhyML 4G SPR (3) 76.3 a 38.1 c 38.1 a 0.847 c 0.847 a
PhyML 4G (SH50) (3) 82.1 a 27.4 b 54.7 a 0.868 b 0.780 a
FastTree (SH50) (3) 86.9 a 35.4 c 51.5 a 0.843 b 0.793 a
PhyML 4G (3) 87.1 a 30.9 c 56.2 a 0.855 c 0.775 a
FastTree (3) 92.2 a 46.0 c 46.2 a 0.814 c 0.814 a
PhyML 1G (SH50) (3) 93.1 a 46.6 c 46.5 a 0.814 c 0.811 a
DendroBLAST (1) 95.1 27.6 67.5 0.867 0.732
PhyML 1G (3) 96.5 b 48.2 c 48.2 a 0.806 c 0.806 a
FastME SPR (2) 119.5 c 59.8 c 59.8 a 0.759 c 0.759 a
FastME (2) 123.7 c 61.9 c 61.9 a 0.751 c 0.751 a
QuickTree log cor. (2) 124.0 c 62.0 c 62.0 a 0.748 c 0.748 c
QuickTree (con50) (2) 128.3 c 28.8 b 99.5 c 0.834 c 0.599 c
QuickTree (2) 161.6 c 80.9 c 80.9 c 0.674 c 0.674 c

Please refer to the legend for table 1 for explanations of abbreviations.