Table 3. Expression profiles for 4 genome-wide significant loci in mice osteoblast gene expression experiments and Likelihood-based Causality Model Selection (LCMS) regulatory network analysis in inbred mice.

Cytogenetic Loci	The closest Genea	Nearby Transcriptsb	Experiment 1c	Experiment 2d	Experiment 3e
			PTH stimulated primary osteoblasts	Differential expression during osteoblast development	LCMS analysis for significant eQTL and trait QTL pairs
			Expression, Direction	P-value	# Traits	Max # pairs
2q11.2	TBC1D8	TBC1D8	0	8.23E-01	0	0
		RPL31	++	n.a.	n.a.	n.a.
18q11.2	OSBPL1A	OSBPL1A	+++ ↓	4.59E-03	1	2
		IMPACT	++	3.17E-03	1	2
1p13.2	RAP1A	RAP1A	++	2.12E-02	3	3
		ADORA3	n.a.	n.a.	0	0
8q24.12	TNFRSF11B*	TNFRSF11B	+++ ↓	7.61E-01	0	0
	COLEC10*	COLEC10	n.a.	1.54E-03	1	2

a SNP locates within a gene.

*: For most significantly associated SNP located on the intergenic regions, the closest nearby gene was listed.

b Transcripts: Transcripts from (1) the closest gene; or (2) genes with target SNP located less than 500K bps on the 5' upstream flanking region.

c PTH stimulated primary osteoblasts: 0: not expressed; ++: expression level > 100 in all 3 replicates; +++: expressed in all 3 replicates and regulated by PTH.

d ANOVA was used to test the differential expression across 7 time points (Day 4, 5, 6, 8, 16, 25 and 30 post-induction) during osteoblast development. Bold: p-value < 0.0083 ( = 0.05/6 available transcripts).

e LCMS analysis: Likelihood-based Causality Model Selection to predict candidate genes causally linked with bone phenotypes. Six bone related traits were tested. For each trait test, at least 2 significant pleiotropy of eQTL and trait QTL pairs was considered evidence for a causally relation to the candidate gene.

c,d,e Results of experiment 1,2 and 3 were obtained from different mice strains and different laboratories.

n.a. Expression level is not available, since transcript is not presented on expression arrays.