A parallel genetic algorithm for single class pattern classification and its application for gene expression profiling in Streptomyces coelicolor

Table 3 Results of the search of eleven secondary metabolic gene clusters of S. coelicolor using the GA algorithm trained with kinetic profiles of RED antibiotic and coelicheline gene clusters (in bold).

cluster	SCO beginning	SCO end	n	RED template	%	Coelicheline template	%
CAD complex	3210	3249	39	28	72	2	5
whiE	5327	5350	5	0	0	0	0
RED	5877	5898	22	NA	NA	2	9
desferioxamines	2782	2785	4	1	25	3	75
coelicheline	489	499	11	3	27	NA	NA
TW95a	5314	5320	7	1	14	0	0
isorenicratein	185	191	7	0	0	0	0
eicosapentoic acid	124	129	6	0	0	0	0
NRPS	6429	6438	9	0	0	0	0
siderophore synthase	5799	5801	3	0	0	0	0
deoxysugar synthase	381	401	21	1	5	12	57

Gene clusters for act, coelibactine, tetrahydroxy naftalene, type I polyketide, chalcone synthase, sesquiterpene, type III fatty acid synthase were not present in the chip data matrix. Geosmine and butyrolactone represented only one gene and were therefore excluded from evaluation. SCO beginning and end represent beginning and end of the gene cluster on the chromosome, where n = number of genes in a gene cluster, RED template refers to the number of genes of a gene cluster identified using RED gene cluster as a training set, and coelicheline template refers to the number of genes of a gene cluster identified using the coelicheline gene cluster as a training set.

ISSN: 1471-2164