To identify functional connections for the identified novel genes with known genes, we performed WGCNA in gene expression profiles generated from PBMs. We used RMA (robust multiarray average) algorithm to correct for the background noise and normalize the expression data with the Expression Console software. The software can be downloaded from the Affymetrix website: http://www.affymetrix.com. The co-expression networks were generated using the WGCNA R package [44]. We extended gene boundaries by 20 kb upstream and downstream of the gene [45]. Any gene containing a SNP in the extended region with a p value < 0.01 for at least one of the two BMD trait (FN-BMD and LS-BMD) and 234 genes which were identified to be associated with BMD by previous studies [12,46,47] are here referred to as the nominally significant GWAS geneset (4796 genes). After excluding the non-expressed genes, we identified 3593 probes representing 3593 genes to construct the co-expression network. Finally, we exported the network of the interesting genes with Topology Overlap Matrix (TOM) value > 0.15. We visualized the networks by Cytoscape [48]. Due to that the TOM
