Figure 4.

Methods for associating rare copy number variants (CNVs) to neurodevelopmental disease. (a) Pathogenicity has been classically associated with a de novo or new mutation model. Pathogenic variants are expected to be strongly selected and the prevalence of these CNVs is essentially maintained by de novo occurrence. (b) Case-control association study to infer pathogenicity for a CNV. Locus-specific CNV frequency is compared in cases and controls under the assumption that the pathogenic CNV is enriched in cases that manifest the disease. This comparison is only valid when both the cohorts are matched for age, sex, and ethnicity and assayed on the comparable CNV detection platform. (c) Sliding window or segment-based approach to identify pathogenic regions in the genome. Such analysis can identify a specific genic region or a locus enriched in cases compared with controls. (d) Size-wise comparison of CNV data as a function of frequency is a good estimate of selective pressure on CNVs. This method provides an estimate of the odds ratio for a particular sized variant. (e) Pathway-based analysis for assessing pathogenicity of the individually rare but collectively common CNVs. This model is generally applicable in the study of complex neuropsychiatric disease wherein related genes are thought to interact in a common neurological pathway. An altered homeostatic state resulting in disease is inferred when two or more genes within the same pathway are disrupted. (f) The global CNV rate and gene disruptions as a function of pathogenic association. The total number of rare CNVs and the number of genes disrupted by deletion or duplication can also be considered for testing pathogenicity. Such a method was recently utilized by Pinto and colleagues in a large-scale study of individuals with autism (98).