Identification of novel DNA repair proteins via primary sequence, secondary structure, and homology

BMC Bioinformatics

Table 8 Correlation between size of a genome and number of DNA repair proteins

ENSEMBL version	SVM threshold	Pearson correlation	p-value	Spearman correlation	p-value
v43	10^-6	0.7165	5.8e-06	0.7269	3.6e-06
v46	10^-6	0.6876	1.9e-05	0.7252	3.9e-06
v48	10^-6	0.7206	4.8e-06	0.748	1.3e-06
v43	0.151	0.6776	2.8e-05	0.7660	5.1e-07
v46	0.151	0.6730	3.3e-05	0.7661	5.0e-07
v48	0.151	0.7000	1.1e-05	0.7974	7.9e-08

The correlation is based on scanning of 31 vetebrate genomes in ENSEMBL using Method P. Scan results for a minimally positive threshold reflect the basic decision nature of the SVM algorithm. A threshold of 0.151 reflects the threshold needed to obtain a maximum true positive rate while allowing no false positives, and hence serves as a lower bound and reliable indicator of the positive correlation between genome size and number of repair proteins. ENSEMBL versions 43, 46, and 48 were released in February, April, and December 2007, respectively.

ISSN: 1471-2105