On genome-wide association studies and their meta-analyses: lessons learned from osteoporosis studies.