Sequencing studies in human genetics: design and interpretation

Key Points The interpretation of next-generation sequencing data is technically and conceptually much more challenging than the data used in genome-wide association studies. Minimizing false-positive signals in sequencing studies depends on careful management of the overall work flow and, in particular, on appropriate statistical criteria used to support claims of significant association. One key feature in the interpretation of sequence data is that most researchers currently distinguish among variants in their prior probabilities of influencing disease, either implicitly or explicitly. Considerable development of appropriate ways to do this, however, is still required. Population genetic, phylogenetic and other data sources can help to establish frameworks for distinguishing among the prior probabilities of variants influencing disease. Although establishing appropriate statistical criteria for interpreting sequence data remains a work in progress, good study designs mandate careful consideration and appropriate correction for the real number of tests that are inherent in any given study design. Interpretation of sequence data should always take into account the narrative potential that is inherent in any human genome, in that all genomes carry many functional and probably deleterious (in an evolutionary sense) rare variants that could be used to argue that the mutations influence traits of interest. Whereas functional characterization of pathogenic mutations is essential in order to derive translational benefits from genetic discoveries, functional characterization should not be used to buttress weak statistical arguments for pathogenicity. In general, with only narrowly defined exceptions, evidence of pathogenicity should come from the genetics alone.