DNA methylation differences at growth related genes correlate with birth weight: a molecular signature linked to developmental origins of adult disease?

Table 8 Each of the two random permutation six-gene models that also had positive R² in the Infinium data set (from Table 7) were combined with each random permutation seven-gene model that achieved an R² higher than the L₁ Infinium model (25 models) for a total of 50, 13 gene models

Data Set	Genes in model	Adjusted R² and stability when tested on Infinium Data	Genes in resulting model	Gene name
GoldenGate gene model which achieved R ² = 0.48 (stability 44/48) on the Infinium Data	BEST1 IMPDH2	0.87 (44/48)	CTTN GMDS	Cortactin GDP-mannose 4,6-dehydratase
	*OSBPL5*		IMPDH2	IMP (inosine 5'-monophosphate) dehydrogenase 2
	*PAX3*		*OSBPL5*	oxysterol binding protein-like 5
	PSMC3		PAX3	paired box 3
	SERPINF1		PSMC3	proteasome (prosome, macropain) 26S subunit, ATPase, 3
Infinium gene model which achieved better R ² than our model	CTTN		*REG1B*	regenerating islet-derived 1 beta
	GMDS		RUVBL1	RuvB-like 1 (E. coli)
	*REG1B*		SERPINF1	serpin peptidase inhibitor, clade F (alpha-2 antiplasmin, pigment epithelium derived factor), member 1
	VPS52		VPS52	vacuolar protein sorting 52 homolog
	RUVBL1
	KIAA241

Only one of the 50 models achieved higher R² than the L₁ 13 gene model. Genes in bold have suggested role in fetal or placental growth and development

ISSN: 1755-8794