Finally, the dTDT should not be limited to dosage probabilities from the inference programs only. As a perspective, the next-generation sequencing data will provide a challenge using the method developed in this paper. Similar methods can be used when pedigree members have a SNP chip, and a subset has sequence data. Despite the discordance and poor replication from our test results, we believe that linkage can help identify regions of interest in conjunction with association testing. Computational inference has helped us reduce the experimental cost in that we may only need to do sequencing on a limited number of family members. Keeping this in mind, we need to implement appropriate selection of the most informative families when we do genotyping. All these future directions shall be promising and have potential to inform the field.
