Although it reduces the probability of producing false negative results, the new method produces a very large number of haplotype blocks, and some blocks will correlate with trait values purely by chance. Raising the significance threshold could reduce the number of correlated genes, but, as discussed above, this increases the chance of producing a false negative result. It has been previously concluded that these “spurious associations” render HBCGM unable to identify a true causative genetic factor [2, 3]. However, HBCGM results are only one component of a comprehensive data analysis package that is used for biomedical trait analysis. Causative genetic candidates have been selected from among the many correlated genes by applying orthogonal criteria [16], such as gene expression and metabolomic [17] or curated biologic data [18], or using the genomic regions delimited by prior QTL analyses [19, 20]. An integrated approach, where HBCGM output is analyzed within the context of multiple ‘omic’ (metabolomic, proteomic, or gene expression) datasets, will become an increasingly important part of 21st century biomedical discovery. This requires a paradigm shift, since it is current practice
