Sigma-2: Multiple sequence alignment of non-coding DNA via an evolutionary model

BMC Bioinformatics

Table 2 Yeast benchmark: effect of parameters

Program	dataset	Matches per base^a	dataset	Matches per base^a	Difference
Sigma-2 (defaults)	orthologous	1.9893	shuffled	0.0031	1.9862
Sigma-2 (no bg model)	orthologous	2.1319	shuffled	0.0133	2.1186
Sigma-2(no tr mat)	orthologous	1.9115	shuffled	0.0046	1.9069
Sigma-2(no bg, no tr)	orthologous	2.3799	shuffled	0.0275	2.3524
Sigma-2(p 0.2)	orthologous	2.4139	shuffled	0.0389	2.3750
Sigma-2(p 0.2, no bg, no tr)	orthologous	2.5615	shuffled	0.0586	2.5029
FSA (defaults)	orthologous	2.7996	shuffled	0.3572	2.4424
FSA (gap5)	orthologous	2.4695	shuffled	0.1465	2.3230
Dialign-TX (defaults)	orthologous	2.9501	shuffled	0.8576	2.0925
Dialign-TX (-l 2)	orthologous	2.7498	shuffled	0.4539	2.2959
Mavid (auto)	orthologous	3.3181	shuffled	2.8248	0.4933
Mavid (yeast tree)	orthologous	3.3393	shuffled	2.8713	0.4680

Performance on the yeast benchmark of a subset of the programs in Table 1 when command-line parameters are varied. In Sigma, "no bg" indicates a background model where each nucleotide is equally probable; "no tr" indicates a "uniform" transition matrix where a nucleotide can mutate to any other nucleotide with equal probability. In FSA, "gap5" indicates the command line option --gapfactor 5. In Dialign, -1 2 is the most stringent setting. Mavid tree options are as described in the text.

ISSN: 1471-2105