Skip to main content
. 2005 Jan 20;33(2):511–518. doi: 10.1093/nar/gki198

Table 4.

Accuracy values of several methods for the PREFAB tests

Method Identity (%) CPU time (s)
0–20 20–40 40–70 70–100 All
G-INS-i 46.75 82.77 96.30 98.60 68.85 16 030
H-INS-i 48.22 83.32 95.83 98.64 69.70 15 060
F-INS-i 48.00 83.35 95.83 98.62 69.61 9007
H-INS-1 46.13a 82.16a 95.42b 98.60 68.27 9910
FFT-NS-i 45.72a 80.95a 93.83b 98.55 67.48 4176
FFT-NS-2 43.19a 79.52a 93.23b 98.47 65.74 930.4
FFT-NS-1 40.90a 77.50a 93.46a 98.59 63.92 666.2
TCoffee 2.02 45.30a 82.36b 95.20 98.62 67.96 973 600
PROBCONS 1.06 45.63b 82.10 95.01b 98.18 67.95 142 200
MUSCLE-i 42.77a 80.43a 95.43 98.28 66.05 13 260
MUSCLE-fast 38.44a 76.65a 93.42a 97.91 62.44 544.1
CLUSTAL W (default) 33.96a 74.14a 93.54b 97.85 59.45 12 970

The highest accuracy value within each percent identity range is in bold letters.

aThe difference from the highest accuracy was found to be significant (P < 0.01) by both the Wilcoxon test and the Friedman test.

bThe Wilcoxon test showed a significant difference but the Friedman test did not. See Table 1 for command-line options for each strategy in MAFFT. Options of other programs are as follows:

TCoffee, default;

PROBCONS, default;

MUSCLE-i, muscle -maxiters 1000;

MUSCLE-fast, muscle -sv -maxiters 1 -diags1 -distance1 kbit20_3;

CLUSTAL W (default), default.