As the costs of deep sequencing of clinical samples continues to come down and the extent of routine coverage increases from individual human coding segments towards whole genomes, it will become increasingly imperative that tools and techniques to predict or determine functional DNA from non-functional DNA keep apace. Here we have demonstrated that a combination of methods based on the assumption of evolutionary sequence conservation being a predictor of function is certainly one plausible approach. While it is also true that enhancers or related elements with similar regulatory potential need not be visibly conserved at the linear DNA alignment level 
, a deeper appreciation of enhancer substructure using cis
-Decoder and related methodologies may well define recognizable commonalities among regulatory enhancers. One of the emerging principals from the analysis of Drosophila
neural enhancers is that the repetitive and palindromic elements in a CSC are often preserved in kind and number, but not orientation or position 
. This leads to a conclusion that it is the type of transcription factor binding sites, but not necessarily their position that may define functionality of enhancers.
Here we have demonstrated that using these types of tools we can identify selected putative cis-regulatory elements and test them for functionality in a convenient animal model using transient transgenesis in zebrafish (see also other examples, 
). The method is estimated to be informative across distantly related vertebrate species in a substantial fraction of cases 
. Although our present case shows that no single test system will be sufficient, we suspect that zebrafish will nonetheless prove to be useful for identifying additional regulatory elements required for ZIC2
expression and additional HPE genes, based on the ancient requirements of forebrain development and patterning. Concurrently, mouse geneticists are working on parallel transient transgenesis approaches that may prove even more useful in translational research of human genetic variation 
. Estimates that 4–6% of the human genome is non-coding sequence with likely regulatory function dictates that this dilemma should remain a priority for both basic scientists and clinicians.
Although we can show that the 3′UTR element is a target of mutation of likely relevant sequence changes among HPE subjects, we have yet to demonstrate the consequences of these base pair alterations. This remains a challenge for the future. Despite its similarity to neural enhancers in Drosophila, our element may well have unappreciated functions. Furthermore, bioinformatic analysis alone does not tell us which of the multiple sequence elements contained within the conservation block are essential, nor which elements are utilized by both species (in the same or similar way), nor which additional functions have evolved due to sequence divergence and adaption. Basic research into these questions will be essential for progress in this area.
The pattern of mutation of our ZIC2
element in this study is entirely analogous to what is seen with the more conventional sequencing of its coding exons. The mutations are rare variants that cannot be readily extrapolated from public databases. In most cases, there will be no information on these variants in extensive public databases, or by comparisons between different diagnostic laboratories. Given this fact, we now conclude that this type of regulatory element be sequenced prospectively in all new cases. Variants of all types either from subjects or controls should be considered for functional testing whenever this is feasible. Several of our subjects were observed to have mutations in more than one HPE risk amplicon (coding and non-coding, see Table S1). This observation is likely only the tip of the iceberg. As the extent of testing of each new subject increases, so will the likelihood of “multiple hits” detected among the battery of tested genes. In the handful of cases that have been adequately examined the observed pattern tends to be of a mutation with a strong attributable risk in conjunction with a normal variant, or one with mildly abnormal function 
. We therefore recommend that the databases of genetic variation ultimately include tests of function of both subject mutations and population variants 
. It is becoming increasingly appreciated that even common polymorphisms can have unappreciated, yet substantial, functional effects to either buffer or enhance the biological consequences of more classical mutations.