Fig. 8

Comparison of the coevolution score found by DeCoTUR with (A) the negative log-p-value of the association found by Coinfinder and (B) the mutual information computed by SpydrPick (without the subsequent ARACNE step). Only coevolution scores that were significant were used. In both cases, there is a general positive correlation between the methods, though they each have different filtering cutoffs. The relationship is less strong with SpydrPick, which is perhaps because many spurious associations were not filtered out by ARACNE (though running the ARACNE step on a dataset of this size removed all the interactions). Note that Mutual Information is an always-positive quantity, so the values that corresponded to negative coevolution scores were artificially labeled as negative