A process for analysis of microarray comparative genomics hybridisation studies for bacterial genomes

BMC Genomics

Table 2 Comparison of the six algorithms, using the microarray hybridization data from the EDL933 sequenced strain. The result of using each algorithm were compared to BLASTN data and are shown below.

	Cut-off	TP	FP	TN	FN	Sensitivity	Specificity	M-Score
	0.50	5202	24	475	43	99.18	95.19	98.83
Naïve	0.33	5211	31	468	34	99.35	93.79	98.87
cut-off	0.25	5218	37	462	27	99.49	92.59	98.89
	0.20	5222	43	456	23	99.56	91.38	98.58
	0.10	5235	127	372	10	99.81	74.55	97.61
GENCOM*		5167	21	478	78	98.31	95.79	98.09
GACK**	EPP = 50	5083	22	477	162	96.91	95.59	96.80
	EPP= 0	5197	24	475	48	99.08	95.19	98.75
Prowollik		5185	23	476	57	98.91	95.39	98.60
MKD***		5207	29	470	38	99.28	94.18	98.84
Mixture	Bimodal	5215	37	462	30	99.42	92.59	98.83
	Trimodal	5207	27	472	38	99.27	94.59	98.86

The number of genes estimated as correctly conserved (True positives, TP), genes identified as conserved but actually are variable (false positives, FP), genes identified as correctly variable (true negatives TN), and genes identified falsely as variable (false negative FN), are given. The sensitivity, specificity, and M-Score are also calculated, where the sensitivity = TP/(TP+FP), specificity = TN/(FP+TN), and M-Score = Sensitivity*prevalence + Specificity*(1-prevalence)
* Institute of Food Research method (GENCOM)
** Genotyping Analysis by Charlie Kim method (GACK)
*** Minimum Kernel Density method (MKD)

ISSN: 1471-2164