A comparative study of S/MAR prediction tools

BMC Bioinformatics

Table 1 Percentage of S/MARs and pseudo-S/MARs predicted by each method by type of sequence

	Positive	Background	Negative	Coding	E. coli
MAR-Finder (rules 1–6)	18.2%	9.4%	0.0%	0.0%	4.5%
MAR-Finder (rules 1–5)	10.3%	9.4%	0.0%	0.0%	1.5%
Duplex destabilisation	13.3%	9.4%	16.4%	15.2%	8.5%
ChrClass	13.9%	10.9%	1.8%	0.6%	0.9%
H-Rule	17.6%	9.7%	0.0%	0.0%	0.0%
Thermodyn	18.8%	9.4%	0.0%	0.0%	0.0%
AT-method	15.2%	9.4%	0.0%	0.0%	0.0%

MAR-Finder (rules 1–6) includes the "AT-richness rules", and MAR-Finder (rules 1–5) excludes this rule. Two or more hits for the same S/MAR counted as one. This Table shows that there is little difference between the results for the positive and background datasets. The ranking of the methods depends on the chosen background discovery rate (see Figure 3).

ISSN: 1471-2105