Nevertheless, the ongoing development of pathway-based tools would benefit from further empirical evaluation of current approaches. For example, a creative meta-analysis might examine how various association metrics affect the likelihood of replication of findings. In addition, testing association methods against well-calibrated positive and negative control datasets might illuminate their relative capabilities. Notably, one study employed multiple pathway analysis algorithms using an extensively-explored Crohn’s disease data set [76]; however, the algorithms chosen were highly-disparate in their null hypotheses and approaches to LD, making it difficult to uniformly compare their results. Alternatively, multi-site collaborations might simultaneously analyze several large data sets using a small number of analytical tools in the same conceptual category; comparisons of the results would advance the underlying science and critically evaluate tools against closely-related options.
