Table 6.

True Positive Summary

	Alignments	Better (count)	Worse (count)	Better (avg. (stdev))	Worse (avg. (stdev))
High Similarity	model trim vs. optimal	10	0	0.018 (0.014)	0
	robust trim vs. optimal	8	0	0.019 (0.014)	0
	frequency trim vs. optimal	11	0	0.015 (0.013)	0
	local trim vs. optimal	15	0	0.024 (0.016)	0
	local trim + model trim vs. local	4	6	0.013 (0.010)	-0.009 (0.003)
	model alignment vs. optimal	8	3	0.024 (0.031)	-0.008 (0.003)
	robust alignment vs. optimal	3	12	0.024 (0.013)	-0.170 (0.170)
	frequency alignment vs. optimal	8	1	0.022 (0.030)	-0.012
Medium Similarity	model trim vs. optimal	14	1	0.056 (0.058)	-0.009
	robust trim vs. optimal	14	0	0.061 (0.056)	0
	frequency trim vs. optimal	13	1	0.053 (0.057)	-0.009
	local trim vs. optimal	14	3	0.098 (0.084)	-0.012 (0.005)
	local trim + model trim vs. local	9	8	0.061 (0.056)	-0.015 (0.009)
	model alignment vs. optimal	10	2	0.039 (0.043)	-0.020 (0.018)
	robust alignment vs. optimal	2	19	0.062 (0.000)	-0.250 (0.140)
	frequency alignment vs. optimal	8	2	0.045 (0.043)	-0.020 (0.018)
Low Similarity	model trim vs. optimal	12	1	0.054 (0.066)	-0.024
	robust trim vs. optimal	17	0	0.051 (0.055)	0
	frequency trim vs. optimal	12	1	0.055 (0.066)	-0.024
	local trim vs. optimal	18	2	0.110 (0.100)	-0.280 (0.380)
	local trim + model trim vs. local	10	3	0.041 (0.065)	-0.013 (0.005)
	model alignment vs. optimal	7	2	0.021 (0.014)	-0.034 (0.022)
	robust alignment vs. optimal	2	16	0.150 (0.050)	-0.240 (0.170)
	frequency alignment vs. optimal	5	3	0.022 (0.016)	-0.029 (0.018)

Numbers shown are the number of sequence pairs, out of 22 in each similarity group, whose percentage of aligned edges that are correct with respect to the Dali alignment are improved or worsened by at least 0.005, along with the average and standard deviation of the magnitude of the difference. In the case of alignments limited to the local alignment boundaries, the Dali alignment is also limited to the local alignment boundaries. Alignments are as described in Table 5.